U.S. patent application number 11/431410 was filed with the patent office on 2009-02-05 for systems and methods for identity-based encryption and related cryptographic techniques.
Invention is credited to Dan Boneh, Matthew Franklin.
Application Number | 20090034714 11/431410 |
Document ID | / |
Family ID | 23209176 |
Filed Date | 2009-02-05 |
United States Patent
Application |
20090034714 |
Kind Code |
A9 |
Boneh; Dan ; et al. |
February 5, 2009 |
Systems and methods for identity-based encryption and related
cryptographic techniques
Abstract
A method and system for encrypting a first piece of information
M to be sent by a sender [100] to a receiver [110] allows both
sender and receiver to compute a secret message key using
identity-based information and a bilinear map. In a one embodiment,
the sender [100] computes an identity-based encryption key from an
identifier ID associated with the receiver [110]. The identifier ID
may include various types of information such as the receiver's
e-mail address, a receiver credential, a message identifier, or a
date. The sender uses a bilinear map and the encryption key to
compute a secret message key g.sub.ID.sup.r, which is then used to
encrypt a message M, producing ciphertext V to be sent from the
sender [100] to the receiver [110] together with an element rP. An
identity-based decryption key d.sub.ID is computed by a private key
generator [120] based on the ID associated with the receiver and a
secret master key s. After obtaining the private decryption key
from the key generator [120], the receiver [110] uses it together
with the element rP and the bilinear map to compute the secret
message key g.sub.ID.sup.r, which is then used to decrypt V and
recover the original message M. According to one embodiment, the
bilinear map is based on a Weil pairing or a Tate pairing defined
on a subgroup of an elliptic curve. Also described are several
applications of the techniques, including key revocation,
credential management, and return receipt notification.
Inventors: |
Boneh; Dan; (Palo Alto,
CA) ; Franklin; Matthew; (Berkeley, CA) |
Correspondence
Address: |
LUMEN PATENT FIRM
2345 YALE STREET
SECOND FLOOR
PALO ALTO
CA
94306
US
|
Prior
Publication: |
|
Document Identifier |
Publication Date |
|
US 20070041583 A1 |
February 22, 2007 |
|
|
Family ID: |
23209176 |
Appl. No.: |
11/431410 |
Filed: |
May 9, 2006 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10218697 |
Aug 13, 2002 |
7113594 |
|
|
11431410 |
May 9, 2006 |
|
|
|
60311946 |
Aug 13, 2001 |
|
|
|
Current U.S.
Class: |
380/28 ;
713/182 |
Current CPC
Class: |
H04L 2209/04 20130101;
H04L 9/3073 20130101; H04L 9/0847 20130101 |
Class at
Publication: |
380/028 ;
713/182 |
International
Class: |
H04L 9/28 20060101
H04L009/28 |
Goverment Interests
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
[0002] The present invention was made with the support of DARPA
contract F30602-99-1-0530. The U.S. Government has certain rights
in the invention.
Claims
1. A method for using identity-based encryption (IBE) to securely
convey data over a communications network from a sender to a
receiver, comprising: at the sender, forming an IBE public key
using a time element; at the sender, obtaining IBE system
parameters; at the sender, encrypting the data using the IBE public
key and the IBE system parameters; at the receiver, receiving the
encrypted data from the sender; at the receiver, obtaining an IBE
private key that corresponds to the IBE public key; and at the
receiver, decrypting the encrypted data using the IBE private
key.
2. The method defined in claim 1 wherein encrypting the data using
the IBE public key and the IBE system parameters comprises
encrypting the data using the IBE public key, the IBE system
parameters, and a bilinear map.
3. The method defined in claim 1 wherein the receiver has
identifying information and wherein forming the IBE public key
comprises forming the IBE public key using the time element and
using the identifying information of the receiver.
4. The method defined in claim 1 wherein the time element includes
a date and wherein forming the IBE public key at the sender
comprises using the date in forming the IBE public key.
5. The method defined in claim 1 wherein the time element includes
a future date and wherein forming the IBE public key at the sender
comprises using the future date in forming the IBE public key.
6. The method defined in claim 1 wherein a private key generator is
connected to the communications network, the method further
comprising: at the receiver, obtaining the IBE private key from the
private key generator over the communications network.
7. A method for using identity-based encryption (IBE) to securely
convey data over a communications network from a sender to a
receiver, wherein the receiver has a credential, comprising: at the
sender, forming an IBE public key using the credential; at the
sender, obtaining IBE system parameters; at the sender, encrypting
the data using the IBE public key and the IBE system parameters; at
the receiver, receiving the encrypted data from the sender; at the
receiver, obtaining an IBE private key that corresponds to the IBE
public key; and at the receiver, decrypting the encrypted data
using the IBE private key.
8. The method defined in claim 7 wherein encrypting the data using
the IBE public key and the IBE system parameters comprises
encrypting the data using the IBE public key, the IBE system
parameters, and a bilinear map.
9. The method defined in claim 7 wherein a private key generator is
connected to the communications network, the method further
comprising: at the receiver, obtaining the IBE private key from the
private key generator over the communications network.
10. The method defined in claim 7 wherein a private key generator
is connected to the communications network, the method further
comprising: at the private key generator, determining whether the
receiver has the credential; and if the private key generator
determines that the receiver has the credential, providing the IBE
private key to the receiver from the private key generator over the
communications network.
11. The method defined in claim 7 wherein the receiver has
identifying information and wherein forming the IBE public key
comprises forming the IBE public key using the identifying
information of the receiver.
12. The method defined in claim 7 wherein the receiver has an
identity and wherein a private key generator is connected to the
communications network, the method further comprising: at the
private key generator, receiving a request for the IBE private key
from the receiver that includes information related to the
credential and that includes the identity of the receiver; at the
private key generator, verifying that the credential belongs to the
receiver; and if the credential is verified as belonging to the
receiver, providing the requested IBE private key to the receiver
from the private key generator.
13. The method defined in claim 7 wherein the credential includes a
security clearance and wherein forming the IBE public key comprises
forming the IBE public key using the security clearance.
14. The method defined in claim 7 wherein the receiver has
identifying information, wherein forming the IBE public key
comprises forming the IBE public key using the identifying
information of the receiver, and wherein encrypting the data using
the IBE public key and the IBE system parameters comprises
encrypting the data using the IBE public key, the IBE system
parameters, and a bilinear map.
15. The method defined in claim 7 wherein the receiver has
identifying information, wherein forming the IBE public key
comprises forming the IBE public key using the identifying
information of the receiver and a time element, and wherein
encrypting the data using the IBE public key and the IBE system
parameters comprises encrypting the data using the IBE public key,
the IBE system parameters, and a bilinear map.
16. An identity-based cryptographic method comprising: at a first
party, forming an identity-based public key using a time element;
at the first party, performing cryptographic operations using the
identity-based public key; at a second party, obtaining an
identity-based private key corresponding to the identity-based
public key; and at the second party, performing cryptographic
operations using the identity-based private key.
17. The identity-based cryptographic method defined in claim 16
wherein performing the cryptographic operations using the
identity-based public key comprises performing the cryptographic
operations using the identity-based public key and a bilinear
map.
18. An identity-based cryptographic method comprising: at a first
party, forming an identity-based public key using a credential of a
second party; at the first party, performing cryptographic
operations using the identity-based public key; at a second party,
obtaining an identity-based private key corresponding to the
identity-based public key; and at the second party, performing
cryptographic operations using the identity-based private key.
19. The identity-based cryptographic method defined in claim 18
wherein performing the cryptographic operations using the
identity-based public key comprises performing the cryptographic
operations using the identity-based public key and a bilinear
map.
20. The identity-based cryptographic method defined in claim 18
further comprising: at the second party, obtaining the
identity-based private key from a private key generator over a
communications network.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation of copending U.S. patent
application Ser. No. 10/218,691 filed Aug. 13, 2002, which claims
the benefit of U.S. provisional application No. 60/311,946, filed
Aug. 13, 2001, both of which are incorporated herein by
reference.
BACKGROUND OF THE INVENTION
[0003] The field of the present invention relates generally to
cryptographic systems.
[0004] Public-key cryptographic systems allow two people to
exchange private and authenticated messages without requiring that
they first have a secure communication channel for sharing private
keys. One of the most widely used public-key cryptosystem is the
RSA cryptosystem disclosed in U.S. Pat. No. 4,405,829. The RSA
cryptosystem is currently deployed in many commercial systems. It
is used by web servers and browsers to secure web traffic, it is
used to ensure privacy and authenticity of e-mail, it is used to
secure remote login sessions, and it is at the heart of electronic
credit-card payment systems. In short, RSA is frequently used in
applications where security of digital data is a concern.
[0005] According to public-key cryptosystems such as the RSA
cryptosystem, each person has a unique pair of keys: a private key
that is a secret and a public key that is widely known. This pair
of keys has two important properties: (1) the private key cannot be
deduced from knowledge of the public key alone, and (2) the two
keys are complementary, i.e., a message encrypted with one key of
the pair can be decrypted only with the complementary key. In these
systems, both the public key and the private key in a pair are
generated together as the output of a key generation algorithm that
takes as input a random seed. Consequently, in these cryptosystems,
people cannot choose a desired public or private key, but must
simply use the keys that are generated for them by a key generation
algorithm. This has the disadvantage that others cannot encrypt
messages to a person until that person generates and publishes a
public key. Another problem with this type of cryptosystem is that
an impostor can publish a public key and claim that it belongs to
someone else. To address this issue, a trusted certificate
authority (CA) is used to authenticate individuals and certify to
others that the individual's public key is authentic.
Unfortunately, this adds complexity to the cryptosystem since a
sender must obtain a certificate for every receiver, and must
obtain a new certificate every time an existing certificate
expires. It also requires receivers to create public keys, publish
them, register certificates with the CA, and renew such
certificates when they expire.
[0006] In 1984 Shamir envisioned a new type of public key
encryption scheme (described in A. Shamir, "Identity-based
cryptosystems and signature schemes", in Advances in
Cryptology--Crypto '84, Lecture Notes in Computer Science, Vol.
196, Springer-Verlag, pp. 47-53, 1984). According to Shamir's
scheme, a person's public key consists of a public identifier,
which may be the person's name and network address, or combination
of name and e-mail address, social security number, street address,
telephone number, or office address. Because the public key is the
person's pre-existing public identifier (ID) rather than a key
produced from a random seed, this kind of public key cryptosystem
is called an identity-based encryption (IBE) scheme. Shamir,
however, did not provide a concrete, practical IBE cryptosystem. In
fact, Shamir argued that existing cryptosystems (such as RSA) could
not be adapted to realize a secure IBE cryptosystem.
[0007] In the years since Shamir proposed his IBE scheme there have
been several attempts to realize an identity-based cryptosystem.
Some proposals require that users not collude. Other proposals
require the private key generator (PKG) to spend an impractically
long time for each private key generation request. Some proposals
require tamper resistant hardware.
[0008] In short, there remains a need for improved cryptographic
methods and systems.
SUMMARY OF THE INVENTION
[0009] According to one embodiment of the invention, a method of
encrypting a first piece of information to be sent by a sender to a
receiver uses an encryption key generated from a second piece of
information. A bilinear map and the encryption key are used to
encrypt at least a portion of the first piece of information to be
sent from the sender to the receiver. The bilinear map may be
symmetric or asymmetric. The bilinear map may be based on a Weil
pairing or a Tate pairing defined on an algebraic group derived
from an elliptic curve. More generally, the bilinear map may be
based on a pairing defined on algebraic varieties.
[0010] According to one embodiment of the invention, encrypting the
portion of the first piece of information can be completed prior to
generating a decryption key corresponding to the encryption
key.
[0011] According to another embodiment of the invention, the second
piece of information is known to the receiver prior to the
generation of a decryption key corresponding to the encryption key.
The second piece of information may comprise a character string
such as an e-mail address, name or other identifier associated with
the receiver, according to different embodiments of the invention.
The second piece of information may also include, according to
various embodiments, an attribute associated with the receiver or
information corresponding to a time or times, such as a date or
series of dates defining one or more time intervals. A decryption
key may be provided based on a time that a request for the
decryption key is received relative to the information
corresponding to a time. According to other embodiments of the
invention, the second piece of information may include a message
identifier, a credential identifier or a message subject
identifier.
[0012] According to another embodiment of the invention, a message
key is generated from the encryption key using a bilinear map, and
a cryptographic hash function is applied to the message key.
[0013] According to another embodiment of the invention, encrypting
the portion of the first piece of information includes generating a
mask from the second piece of information using a bilinear map. The
mask is applied to the portion of the second piece of
information.
[0014] An embodiment of the invention is directed to a method of
decrypting ciphertext which has been encrypted by a sender using an
identity-based encryption key associated with a receiver. A
decryption key derived from the encryption key is obtained. At
least a portion of the ciphertext is decrypted using a bilinear map
and the decryption key. The bilinear map may be symmetric or
asymmetric. The bilinear map may be based on a Weil pairing or a
Tate pairing defined on an algebraic group derived from an elliptic
curve.
[0015] According to another embodiment of the invention, the
ciphertext is obtained prior to creating the decryption key.
According to another embodiment of the invention, the first piece
of information is known to the receiver prior to obtaining the
ciphertext and prior to obtaining the decryption key. The
decryption key may be obtained by sending a request to a private
key generator, including information sent together with the
ciphertext.
[0016] An embodiment of the invention is directed to a method of
generating a decryption key corresponding to an encryption key. An
algebraic group, a group action, and a master key are provided. The
encryption key is generated based on a first piece of information.
The decryption key is generated based on the group action, the
master key and the encryption key. According to one embodiment of
the invention, the group action is capable of being calculated in
polynomial time. According to another aspect of the invention,
generation of the decryption key in the absence of the master key
would require greater than polynomial time.
[0017] Another embodiment of the invention is directed to a method
of providing system parameters for a cryptographic system.
Algebraic groups .sub.1 and .sub.2 having an order q are provided,
together with associated group actions. In addition, a bilinear map
is provided that maps pairs of points in .sub.1 to points in
.sub.2. In another embodiment, a system parameter representing a
member P of G.sub.1, and a system parameter representing a member
P.sub.pub of G.sub.1 are provided, where P.sub.pub is based on the
group action of a master key s applied to P. According to other
embodiments of the invention, a system parameter representing a set
of one or more hash functions H.sub.1, H.sub.2, H.sub.3, or H.sub.4
are provided. According to another embodiment of the invention, a
system parameter representing a size n of a message space is
provided.
[0018] According to another embodiment of the invention, the
bilinear map may be asymmetric or symmetric. In another embodiment
the bilinear map is based on a Weil pairing or a Tate pairing
defined on a portion of an elliptic curve.
[0019] According to another embodiment of the invention, the
algebraic group G.sub.1 is defined by an elliptic curve defined
over a field of order p and the order q is less than the order p.
According to another aspect of the invention, the length of p is at
least 1024 bits and the length of q is no greater than 160
bits.
[0020] Another embodiment of the invention is directed to a method
for managing cryptographic communication including generating
shares of a master key. The shares are stored in separate systems.
A request from a receiver to obtain a private key is responded to
in the separate systems by generating from the respective shares of
the master key, corresponding respective shares of the private key.
The receiver constructs the private key from the shares of the
private key, where the private key corresponds to identifying
information of the receiver.
[0021] Another embodiment of the invention is directed to a method
for communicating between a sender and a receiver. A message to be
sent from the sender to the receiver is encrypted, and the message
is sent from the sender to the receiver. A request for a decryption
key is received from the receiver of the message. After receiving
the request for the decryption key, information indicating that the
receiver has received the message is generated, and the decryption
key is provided to the receiver. According to an embodiment of the
invention, a return address of the sender is included in the
message, and an acknowledgment that the message has been received
is sent to the return address. According to another aspect of the
invention, an identification of the message is included in an
acknowledgment and the acknowledgment is sent to the sender.
According to another aspect of the invention, the encryption key is
derived based on a return address of the sender.
[0022] Another embodiment of the invention is directed to a method
for communicating between a sender and a receiver having a
credential. Identifying information of the receiver is obtained. A
credential required for the receiver to gain a decryption key is
specified, and an encryption key is derived from the identifying
information of the receiver and the credential. A message to be
sent from the sender to the receiver is encrypted using the
encryption key and a bilinear map, and the message is sent from the
sender to the receiver. A request for a decryption key is received
from the receiver of the message. It is determined whether the
receiver has the credential, and if the receiver has the
credential, the decryption key is provided to the receiver. The
receiver then may use the decryption key and the bilinear map to
decrypt the message.
[0023] Another embodiment of the invention is directed to a method
of communicating between a sender and a receiver involving storing
a decryption key on a target system. Sets of decryption keys
associated with times messages may be decrypted are derived, and
the decryption keys are stored on the target system. An encryption
key is derived from a string associated with a time a message is to
be decrypted. A message is encrypted using the encryption key. The
message is received on the target system, and the message is
decrypted using a bilinear map and the corresponding decryption
key.
[0024] Another embodiment of the invention is directed to a method
of communicating between a sender and receiver involving entities
having different responsibilities. A set of decryption keys is
derived from a master key and a set of strings associated with
different responsibilities. The decryption keys are provided to
entities having the respective responsibilities. An encryption key
is derived from a string associated with one of the different
responsibilities. A message to be sent from the sender to the
receiver is encrypted using the encryption key and a bilinear map.
An entity having a particular responsibility receives the message
and decrypts the message using the respective decryption key and
the bilinear map. According to one embodiment of the invention, the
string corresponding to the particular responsibility comprises a
subject line of an e-mail.
BRIEF DESCRIPTION OF THE DRAWING FIGURES
[0025] FIG. 1 is a block diagram illustrating a cryptosystem
according to an embodiment of the invention, showing steps taken by
a sender, a receiver, and a private key generator (PKG), and
information communicated between them.
[0026] FIG. 2 is a block diagram illustrating steps performed by a
PKG when generating a private key according to an embodiment of the
invention.
[0027] FIG. 3 is a block diagram illustrating steps performed by a
sender when computing a secret message key and using it to encrypt
a message intended for a receiver according to an embodiment of the
invention.
[0028] FIG. 4 is a block diagram illustrating steps performed by a
receiver when computing a secret message key and using it to
decrypt ciphertext received from a sender according to an
embodiment of the invention.
[0029] FIG. 5 is a block diagram illustrating a distributed PKG,
according to an embodiment of the invention.
[0030] FIG. 6 is a block diagram illustrating elements in a
cryptosystem with escrow decryption capability according to an
embodiment of the invention.
[0031] FIG. 7 is a block diagram illustrating steps performed by a
sender when encrypting messages in an ElGamal cryptosystem with
escrow decryption capability according to an embodiment of the
invention.
[0032] FIG. 8 is a block diagram illustrating steps performed by a
receiver when decrypting messages in an ElGamal cryptosystem with
escrow decryption capability according to an embodiment of the
invention.
[0033] FIG. 9 is a block diagram illustrating steps performed by an
escrow when decrypting messages in an ElGamal cryptosystem with
escrow decryption capability according to an alternate embodiment
of the invention.
[0034] FIG. 10 is a block diagram illustrating a system for
managing credentials in an identity based encryption system
according to an embodiment of the invention.
[0035] FIG. 11 is a block diagram illustrating a system with key
delegation according to an embodiment of the invention.
[0036] FIG. 12 is a block diagram illustrating an encryption system
with return receipt according to an embodiment of the
invention.
DETAILED DESCRIPTION OF THE INVENTION
[0037] The following description provides details of several
exemplary embodiments of the cryptographic techniques of the
present invention, as well as a technical discussion of the
security of the system.
Overview
[0038] As is normally the case with modern cryptosystems, the
techniques of the present invention are generally implemented on
computers connected by a communication medium. Although typically
the computers are connected by the Internet or another computer
network, any communication medium may be used.
[0039] One embodiment of the invention comprises an identity-based
encryption system that uses a secret message key derived from
identity-based information. The message key may be used by a sender
to encrypt a message, and by a receiver to decrypt the message. The
secret message key is computed by the sender from an identity-based
public key of the receiver. The same message key may be computed by
the receiver from the receiver's private key, which is derived from
the receiver's identity-based public key. Both sender and receiver
compute the same secret key using a bilinear map. For example, in
one embodiment, an asymmetric or symmetric bilinear map :
.sub.0.times..sub.1.fwdarw..sub.2 is used where .sub.0, .sub.1,
.sub.2 are (not necessarily distinct) algebraic groups. In the case
where .sub.0 is equal to .sub.1, we say the bilinear map is
symmetric and often denote it as :
.sub.1.times..sub.1.fwdarw..sub.2. A bilinear map that is
non-degenerate and efficiently computable will be referred to as an
admissible map. It is preferable in some embodiments of the
invention that the bilinear map be admissible.
[0040] The convention throughout this description will be to denote
the group operations of .sub.0 and .sub.1 by addition, and the
group operation of .sub.2 by multiplication. For a group of prime
order we use * to denote the set *=\{O} where O is the identity
element in the group . The set of binary strings of arbitrary
length is denoted by {0, 1}*. We use .sub.q to denote the group {0,
. . . , q-1} under addition modulo q, and we use .sup.+ to denote
the set of positive integers. We note that there is a natural group
action of .sub.q on given by repeated addition, and we denote the
result of the action of an element .alpha..epsilon..sub.q on an
element P.epsilon. by aP.
[0041] According to another embodiment of the invention, a certain
variant (involving the map ) of the computational Diffie-Hellman
problem is hard. In one implementation the map is admissible and
the orders of .sub.0, .sub.1, .sub.2 have a very large prime factor
q. The orders of .sub.0, .sub.1 and .sub.2 may be equal to each
other. Without loss of generality, the following description
assumes for simplicity that the orders of .sub.0, .sub.1, and
.sub.2 are all of prime order q.
[0042] In an exemplary embodiment, an admissible map :
.sub.0.times..sub.1.fwdarw..sub.2 is used to realize an
identity-based cryptosystem, as follows. To encrypt a message, a
sender uses a public key Q.sub.ID.epsilon..sub.1 associated with a
public identifier ID for the intended receiver. To decrypt the
encrypted message, the receiver uses a complementary private key
d.sub.ID.epsilon..sub.1. The private key is computed from the
public key Q.sub.ID, a secret master key s.epsilon.*.sub.q, and a
group action of *.sub.q on .sub.1. In one embodiment, for example,
d.sub.ID=sQ.sub.ID. Since the secret master key s is known only by
a trusted PKG, users normally cannot themselves compute private
keys. To obtain a private key, an individual may obtain it from the
PKG, preferably after being authenticated. At any time, however,
anyone can compute the public key Q.sub.ID associated with any
public identifier ID even before the corresponding private key has
been determined. For example, in one embodiment the public key
Q.sub.ID may be obtained by (1) using a conventional character
encoding scheme to map the public identifier ID to a corresponding
binary string in {0, 1}*, and (2) using a hash function H.sub.1:
{0, 1}*.fwdarw.*.sub.1 to hash the binary string to the element
Q.sub.ID of *.sub.1, where the order of Q.sub.ID is q.
[0043] In this embodiment, a message intended for a receiver with
public identifier ID may be encrypted and decrypted as follows. The
admissible map may be used by the sender to determine a secret
message key. Although the sender and receiver do not share all the
same information, using the fact that the map is bilinear, they can
use different information to compute the same message key. Since
each uses information that is private, the message key is a
secret.
[0044] To illustrate how this approach may be implemented, suppose
that the sender has knowledge of elements P and sP in .sub.1. In
one embodiment, for example, elements P and P.sub.pub=sP in .sub.1
are published system parameters. Now further suppose the sender
privately selects a random r.epsilon.*.sub.q, and uses the
receiver's identity-based public key Q.sub.ID to compute
g.sub.ID.sup.r= (rQ.sub.ID, sP). The element g.sub.ID.sup.r is an
identity-based secret which the sender may use as a secret message
key to perform identity-based encryption of a message to the
receiver. The sender may then send an encrypted message together
with rP to the receiver. The receiver then receives rP and uses it
together with the private key sQ.sub.ID to compute the secret
message key g.sub.ID.sup.r= (sQ.sub.ID, rP). This secret message
key is equal to the secret message key computed by the sender
because of the bilinearity of the map. This computed element
g.sub.ID.sup.r.epsilon..sub.2 is thus an identity-based secret of
the sender which the receiver may compute using the element rP and
the private key sQ.sub.ID. This secret may be used as a message key
for cryptographic communication between the sender and
receiver.
[0045] Note that the PKG also knows the receiver's private key, so
can also compute the secret message key and decrypt the message.
The sender, receiver and PKG all have sufficient information to
compute the secret message key. No other entity, however, normally
has knowledge of the sender's secret r or the receiver's secret
sQ.sub.ID. The security of this embodiment is related to the
difficulty of computing the secret message key, which is based upon
a combination of r, s, and Q.sub.ID using a bilinear map, without
knowledge of r or knowledge of sQ.sub.ID.
[0046] In one embodiment, the message key g.sub.ID.sup.r is used to
determine a mask which is used to encrypt and decrypt the bits of
the message using an XOR operation (denoted by `.sym.`).
Specifically, the ciphertext V of a message M is produced by
computing V=M.sym.H.sub.2(g.sub.ID.sup.r), where H.sub.2:
.sub.2.fwdarw.{0, 1}.sup.n is a hash function, and n is the bit
length of the message. Conversely, the message M is recovered from
the ciphertext V by computing M=V.sym.H.sub.2(g.sub.ID.sup.r).
[0047] In another embodiment, the one-way encryption scheme
outlined above is made more secure by converting it into a chosen
ciphertext secure system. In one embodiment of the invention, for
example, a general technique of Fujisaki-Okamoto is used.
[0048] In another embodiment, the master key is broken into
components s.sub.i distributed among several private key generators
in a distributed PKG. For a given user with a public key Q.sub.ID
based on an identifier ID, each of these private key generators in
the distributed PKG computes a private key portion d.sub.i using Q
and its portion s.sub.i of the master key. These private key
portions can be combined by the user and used as a single private
key d.sub.ID to decrypt messages encrypted with Q.sub.ID.
[0049] In another embodiment, an ElGamal encryption scheme is
provided with built-in key escrow, i.e., where one global escrow
key can decrypt ciphertexts encrypted under any public key. In this
embodiment, the exemplary system described above is adapted as
follows. Suppose that the receiver also has knowledge of elements P
and sP. Rather than obtaining a private key from the PKG, the
receiver generates a public/private key pair by selecting a random
x.epsilon.*.sub.q, computing xP using a group action, and
publishing a public key based on the result of the computation. In
one embodiment, the public key is xP and the complementary private
key is d=x(sP). (Thus, xP plays the role of Q.sub.ID, and
d=x(sP)=s(xP) plays the role of d.sub.ID=sQ.sub.ID.) To encrypt a
message to the receiver, the sender as before selects a random r
and sends rP to the receiver. Then the receiver knows the pair (rP,
x(sP)), where x(sP)=d is a secret, while the sender knows the pair
(sP, r(xP)), where r(xP) is a secret. Thus, the sender and receiver
both can compute g=e(rP, x(sP))=e(sP, r(xP)), where the second
equality follows from the bilinearity of . This secret, however,
can also be determined from knowledge of the master key s. Using
the element rP from the sender, the receiver's public key xP, and
the master key s, the message key can be computed by evaluating g=
(rP, s(xP)). It should be noted that this embodiment makes use of a
symmetric bilinear map : .sub.1.times..sub.1.fwdarw..sub.2.
[0050] In several embodiments of the invention, .sub.1 is a
subgroup of an elliptic curve, and an admissible map is constructed
from the Weil pairing (or Tate pairing) on the elliptic curve.
(Recall that, by definition, a subgroup is not necessarily smaller
than the group, i.e., .sub.1 may be the entire elliptic curve).
More generally, .sub.1 may be an abelian variety and an admissible
pairing of its elements. In embodiments using a map :
.sub.0.times..sub.1.fwdarw.G.sub.2 where .sub.0 and .sub.1 are
distinct, .sub.0 also may be a subgroup of an elliptic curve, or
more generally, an abelian variety.
[0051] In other embodiments, various novel applications of
identity-based encryption are provided. New and useful applications
of IBE systems are possible by using other types of public
identifiers, or enhanced public identifiers. For example, the
public identifier ID is not limited to an identifier associated
with an individual person, but may be an identifier associated with
any type of entity including not just individuals but also
organizations, governmental agencies, corporations and the like. It
should also be noted that individual identities forming a group may
be naturally combined to produce a joint identity for the group
with a corresponding group private key. The group's private key
need not be issued by a PKG, but is simply the combination of the
separate private keys of the entities composing the group. It
should be noted that the basic ID specifying the identity of an
entity is not limited to the name, e-mail address, address, or
social security number of an entity, but could also include other
types of information such as domain names, URLs, 9-digit zip codes,
tax identification numbers, and so on. In many applications, the
public identifier ID will contain some character string known to
the public to be uniquely associated with a particular entity or
collection of entities. In general, however, the public identifier
ID can be any arbitrary character string or other arbitrary
information.
[0052] Various useful applications of IBE make use of enhanced
public identifiers. An enhanced identifier may comprise a type of
identifier that contains information not necessarily limited to
information specifying the identity of a particular entity. For
example, an ID can contain a credential descriptor such as a
license number, official title, or security clearance associated
with an entity. An agency can then manage the credentials by
providing private keys only to entities it certifies. In one
exemplary embodiment, an ID can contain a property descriptor such
as a serial number, vehicle identification number, patent number,
or the like. An agency responsible for registering property owners
and authenticating owners can manage property registration by
providing private keys only to entities that it registers as true
owners. More generally, an association between two or more things
can be managed by including identifiers for them in an ID. The PKG
then acts as the management authority for the associations between
things.
[0053] Another type of enhanced ID is an identifier that includes a
time, a time interval, or a set of time intervals. A private key
for such an identifier can then be constructed to automatically
expire at a certain time, to automatically activate only after a
certain time, or to be valid only for one or more specified time
intervals. This technique can be combined with the credential and
ownership management to control the time of activation and/or
expiration.
[0054] From the above examples, it is evident that an
identity-based encryption systems according to the present
invention are not limited to any particular type of identifier.
Thus, the term `identity-based` should be understood in general
terms as indicating that any arbitrary character string or other
arbitrary information may be used as a basis.
[0055] According to another embodiment, an IBE system allows the
delegation of decryption capabilities. An entity can set up its own
IBE system with its own secret master key, and assume the role of
PKG for this IBE system. Because the entity has the master key, it
can issue keys to delegate decryption capabilities to others. For
example, if the entity is a corporation, the employees can obtain
private keys from the corporate PKG. Individuals can be issued
private keys matching their names, titles, duties, projects, cases,
or any other task-related identifier. In another example, an
individual can issue to a laptop private keys that are valid only
for the duration of a business trip. If the laptop is lost or
stolen, only the keys for that time period are compromised. The
master key, which remained at home, is uncompromised.
[0056] It should also be pointed out that the medium of
communication need not be limited to e-mail or the Internet, but
could include any communication medium such as printed
publications, digital storage media, radio broadcasting, wireless
communications, and so on.
Definitions
Identity-Based Encryption. An exemplary embodiment of an
identity-based encryption system and method E uses four randomized
algorithms: Setup, Extract, Encrypt, Decrypt:
[0057] Setup: Given a security parameter k, return params (system
parameters) and master-key. The system parameters include a
description of a finite message space , and a description of a
finite ciphertext space . Normally, the system parameters will be
publicly known, while the master-key will be known only to a
Private Key Generator (PKG). [0058] Extract: takes as input params,
master-key, and an arbitrary ID.epsilon.{0, 1}*, and returns a
private key d. Here ID is an arbitrary string that will be used as
a public key, and d is the corresponding private decryption key.
The Extract algorithm extracts a private key from the given public
key. Because the extraction requires the master-key, it is normally
performed by the PKG. [0059] Encrypt: takes as input params, ID,
and M.epsilon.. It returns a ciphertext C.epsilon.. [0060] Decrypt:
takes as input params, C.epsilon., and a private key d. It return
M.epsilon.. According to an embodiment of the invention, these
algorithms satisfy the standard consistency constraint that ensures
decryption will faithfully recover any encrypted message. More
specifically, when d is the private key generated by algorithm
Extract when it is given ID as the public key, then
.A-inverted.M.epsilon.: Decrypt(params, C, d)=M where
C=Encrypt(params, ID, M).
[0061] In an identity-based cryptosystem according to an embodiment
of the invention, the above algorithms are used together as
illustrated in FIG. 1. A sender 100 uses Encrypt, a receiver 110
uses Decrypt, and a PKG 120 uses Setup and Extract. To send a
message M to receiver 110, the sender 100 obtains an ID of the
receiver (e.g., the receiver's e-mail address) and combines it with
a randomly selected integer r to compute a secret message key
g.sub.ID.sup.r. The element rP is sent to receiver 110 who combines
it with a private key d.sub.ID to determine the same message key
g.sub.ID.sup.r. Because the sender and receiver share the secret
message key, a message encrypted with the key by the sender can be
decrypted by the receiver. In particular, the sender encrypts M
with the message key to produce ciphertext V which is communicated
with rP to the receiver. The receiver then uses the secret message
key to decrypt the ciphertext to recover the original message. In
order to decrypt messages, however, the receiver 110 must first
obtain the private key d.sub.ID from the PKG 120. After the PKG
authenticates the identity of the receiver, it provides the
receiver with the private key corresponding to the receiver's ID.
(Note that, in this embodiment, the PKG can compute any private key
in the system, and can thus decrypt any message to any user in the
system.)
[0062] Chosen ciphertext security. Chosen ciphertext security
(IND-CCA) is the standard acceptable notion of security for a
public key encryption scheme. An embodiment of an identity-based
encryption system and method may be implemented to satisfy this
strong notion of security. Additionally, the selected level of
chosen ciphertext security may be strengthened a bit. The reason is
that when an adversary attacks a public key ID in an identity-based
system, the adversary might already possess the private keys of
users ID.sub.1, . . . , ID.sub.n of her choice. In an embodiment of
the invention, the system remains secure under such an attack. That
is, the system remains secure even when the adversary can obtain
the private key associated with any identity ID.sub.i of her choice
(other than the public key ID being attacked). We refer to such
queries as private key extraction queries. The system of this
embodiment also remains secure even though the adversary is
challenged on a public key ID of her choice (as opposed to a random
public key).
[0063] We say that an embodiment of an identity-based encryption
system or method is semantically secure against an adaptive chosen
ciphertext attack (IND-ID-CCA) if no polynomially bounded adversary
A has a non-negligible advantage against the Challenger in the
following IND-ID-CCA game: [0064] Setup: The challenger takes a
security parameter k and runs the Setup algorithm. It gives the
adversary the resulting system parameters params. It keeps the
master-key to itself. [0065] Phase 1: The adversary issues queries
q.sub.1, . . . , q.sub.m where query q.sub.i is one of: [0066]
Extraction query <ID.sub.i>. The challenger responds by
running algorithm Extract to generate the private key d.sub.i
corresponding to the public key <ID.sub.i>. It sends d.sub.i
to the adversary. [0067] Decryption query <ID.sub.i,
C.sub.i>. The challenger responds by running algorithm Extract
to generate the private key d.sub.i corresponding to ID.sub.i. It
then runs algorithm Decrypt to decrypt the ciphertext C.sub.i using
the private key d.sub.i. It sends the resulting plain-text to the
adversary. [0068] These queries may be asked adaptively, that is,
each query q.sub.i may depend on the replies to q.sub.1, . . . ,
q.sub.i-1. [0069] Challenge: Once the adversary decides that Phase
1 is over it outputs two equal length plain-texts M.sub.0,
M.sub.1.epsilon. and an identity ID on which it wishes to be
challenged. The only constraint is that ID did not appear in any
private key extraction query in Phase 1. [0070] The challenger
picks a random bit b.epsilon.{0, 1} and sets C=Encrypt(params, ID,
M.sub.b). It sends C as the challenge to the adversary. [0071]
Phase 2: The adversary issues more queries q.sub.m+1, . . . ,
q.sub.n where query q.sub.i is one of: [0072] Extraction query
<ID.sub.i> where ID.sub.i.noteq.ID. Challenger responds as in
Phase 1. [0073] Decryption query <ID.sub.i,
C.sub.i>.noteq.<ID, C>. Challenger responds as in Phase 1.
[0074] These queries may be asked adaptively as in Phase 1. [0075]
Guess: Finally, the adversary outputs a guess b'.epsilon.{0, 1}.
The adversary wins the game if b=b'. [0076] We refer to such an
adversary as an IND-ID-CCA adversary. We define adversary 's
advantage in attacking the scheme as the following function of the
security parameter k (k is given as input to the challenger): Adv ,
.function. ( k ) = Pr .function. [ b = b ' ] - 1 2 . ##EQU1##
[0077] The probability is over the random bits used by the
challenger and the adversary. Using the IND-ID-CCA game we can
define chosen ciphertext security for IBE schemes. As usual, we say
that a function g:.fwdarw. is negligible if g(k) is smaller than
1/f(k) for any polynomial f. Definition 1 We say that an IBE system
is semantically secure against an adaptive chosen ciphertext attack
if for any polynomial time IND-ID-CCA adversary the function
Adv.sub.,(k) is negligible. As shorthand, we say that is IND-ID-CCA
secure.
[0078] Note that the standard definition of chosen ciphertext
security (IND-CCA) is the same as above except that there are no
private key extraction queries and the adversary is challenged on a
random public key (rather than a public key of her choice). Private
key extraction queries are related to the definition of chosen
ciphertext security in the multiuser settings. After all, our
definition involves multiple public keys belonging to multiple
users. A multiuser IND-CCA may be reducible to single user IND-CCA
using a standard hybrid argument. This does not hold in the
identity-based settings, IND-ID-CCA, since the adversary gets to
choose which public keys to corrupt during the attack. To emphasize
the importance of private key extraction queries we note that one
implementation of the disclosed IBE system can be modified (by
removing one of the hash functions) into a system which has chosen
ciphertext security when private extraction queries are disallowed.
However, the implementation is insecure when extraction queries are
allowed.
[0079] Semantically secure identity based encryption. The proof of
security for an implementation of our IBE system makes use of a
weaker notion of security known as semantic security (also known as
semantic security against a chosen plain-text attack). Semantic
security is similar to chosen ciphertext security (IND-ID-CCA)
except that the adversary is more limited; it cannot issue
decryption queries while attacking the challenge public key. For a
standard public key system (not an identity based system) semantic
security is defined using the following game: (1) the adversary is
given a random public key generated by the challenger, (2) the
adversary outputs two equal length messages M.sub.0 and M.sub.1 and
receives the encryption of M.sub.b from the challenger where b is
chosen at random in {0, 1}, (3) the adversary outputs b' and wins
the game if b=b'. The public key system is said to be semantically
secure if no polynomial time adversary can win the game with a
non-negligible advantage. As shorthand we say that a semantically
secure public key system is IND-CPA secure. Semantic security
captures our intuition that given a ciphertext the adversary learns
nothing about the corresponding plain-text.
[0080] To define semantic security for identity based systems
(denoted IND-ID-CPA) we strengthen the standard definition by
allowing the adversary to issue chosen private key extraction
queries. Similarly, the adversary is challenged on a public key ID
of her choice. We define semantic security for identity based
encryption schemes using an IND-ID-CPA game. The game is identical
to the IND-ID-CCA game defined above except that the adversary
cannot make any decryption queries. The adversary can only make
private key extraction queries. We say that an identity-based
encryption scheme is semantically secure (IND-ID-CPA) if no
polynomially bounded adversary has a non-negligible advantage
against the Challenger in the following IND-ID-CPA game: [0081]
Setup: The challenger takes a security parameter k and runs the
Setup algorithm. It gives the adversary the resulting system
parameters params. It keeps the master-key to itself. [0082] Phase
1: The adversary issues private key extraction queries ID.sub.1, .
. . , ID.sub.m. The challenger responds by running algorithm
Extract to generate the private key d.sub.i corresponding to the
public key ID.sub.i. It sends d.sub.i to the adversary. These
queries may be asked adaptively. [0083] Challenge: Once the
adversary decides that Phase 1 is over it outputs two equal length
plain-texts M.sub.0, M.sub.1.epsilon. and a public key ID on which
it wishes to be challenged. The only constraint is that ID did not
appear in any private key extraction query in Phase 1. The
challenger picks a random bit b.epsilon.{0, 1} and sets
C=Encrypt(params, ID, M.sub.b). It sends C as the challenge to the
adversary. [0084] Phase 2: The adversary issues more extraction
queries ID.sub.m+1, . . . , ID.sub.n. The only constraint is that
ID.sub.i.noteq.ID. The challenger responds as in Phase 1. [0085]
Guess: Finally, the adversary outputs a guess b'.epsilon.{0, 1}.
The adversary wins the game if b=b'. [0086] We refer to such an
adversary as an IND-ID-CPA adversary. As we did above, the
advantage of an IND-ID-CPA adversary against the scheme is the
following function of the security parameter k: Adv , .function. (
k ) = Pr .function. [ b = b ' ] - 1 2 . ##EQU2## [0087] The
probability is over the random bits used by the challenger and the
adversary. Definition 2 We say that the IBE system is semantically
secure if for any polynomial time IND-ID-CPA adversary the function
Adv.sub.,(k) is negligible. As shorthand, we say that is IND-ID-CPA
secure. One way identity-based encryption. One can define an even
weaker notion of security called one-way encryption (OWE). Roughly
speaking, a public key encryption scheme is a one-way encryption if
given the encryption of a random plain-text the adversary cannot
produce the plain-text in its entirety. One-way encryption is a
weak notion of security since there is nothing preventing the
adversary from, say, learning half the bits of the plaintext.
Hence, one-way encryption schemes do not generally provide secure
encryption. In the random oracle model one-way encryption schemes
can be used for encrypting session-keys (the session-key is taken
to be the hash of the plain-text). We note that one can extend the
notion of one-way encryption to identity based systems by adding
private key extraction queries to the definition. We do not give
the full definition here since we use semantic security as the
weakest notion of security. Bilinear Maps and the Bilinear
Diffie-Hellman Assumption
[0088] One embodiment of the invention is directed to an IBE system
that makes use of a map : .sub.1.times..sub.1.fwdarw..sub.2 between
groups .sub.1 and .sub.2 of order q for some large prime q. A map
may be called an admissible map if it satisfies the following
properties: [0089] 1. Bilinear: The map :
.sub.1.times..sub.1.fwdarw..sub.2 satisfies (aP, bQ)= (P, Q).sup.ab
for all (P, Q).epsilon..sub.1.times..sub.1 and all a, b.epsilon..
[0090] 2. Non-degenerate: The map does not send all pairs in
.sub.1.times..sub.1 to the identity in .sub.2. Observe that since
.sub.1, .sub.1, .sub.2 are groups of prime order this implies that
if .sub.1=.sub.1 and P is a generator of .sub.1=.sub.1 then (P, P)
is a generator of .sub.2. [0091] 3. Computable: There is an
efficient algorithm to compute (P, Q) for any (P,
Q).epsilon..sub.1.times..sub.1. Although many of the embodiments
are described with reference to a map :
.sub.1.times..sub.1.fwdarw..sub.2, this is only a specific case of
bilinear maps used in embodiments of the invention. More generally,
maps : .sub.0.times..sub.1.fwdarw..sub.2 may be used, where .sub.0
and .sub.1 may be distinct. For simplicity of description, however,
the description of many of the embodiments focuses primarily on the
case where .sub.1 and .sub.1 are the same, and both groups are then
denoted .sub.1. Below we present a detailed exemplary embodiment
using groups .sub.1, .sub.2 and an admissible map between them. In
this exemplary embodiment, the group .sub.1 is a subgroup of the
additive group of points of an elliptic curve E/.sub.p, and the
group .sub.2 is a subgroup of the multiplicative group of a finite
field *.sub.p.sub.2. As we will see below in the detailed example
of an IBE system, the Weil pairing (which is not itself an
admissible map) can be used to construct an admissible map between
these two groups.
[0092] The existence of the admissible map :
.sub.1.times..sub.1.fwdarw..sub.2 as above has two direct
implications to these groups. [0093] The MOV reduction: The
discrete log problem in .sub.1 is no harder than the discrete log
problem in .sub.2. To see this, let P, Q.epsilon..sub.1 be an
instance of the discrete log problem in .sub.1 where both P, Q have
order q. We wish to find an a.epsilon..sub.q such that Q=aP. Let g=
(P, P) and h= (Q, P). Then, by bilinearity of we know that
h=g.sup.a. By non-degeneracy of both g, h have order q in .sub.2.
Hence, we reduced the discrete log problem in .sub.1 to a discrete
log problem in .sub.2. It follows that for discrete log to be hard
in .sub.1 we must choose our security parameter so that discrete
log is hard in .sub.2. [0094] Decision Diffie-Hellman is Easy: The
Decision Diffie-Hellman problem (DDH) in .sub.1 is the problem of
distinguishing between the distributions <P, aP, bP, abP> and
<P, aP, bP, cP> where a, b, c are random in .sub.q and P is
random in .sub.1. To see that DDH in .sub.1 is easy, observe that
given P, aP, bP, cP.epsilon.*.sub.1 we have c=ab mod q (P, cP)=
(aP, bP). [0095] The Computational Diffie-Hellman problem (CDH) in
.sub.1 can still be hard (CDH in .sub.1 is to find abP given random
<P, aP, bP>). Exemplary embodiments may use mappings :
.sub.1.times..sub.1.fwdarw..sub.2 where CDH in .sub.1 is believed
to be hard even though DDH in .sub.1 is easy. The Bilinear
Diffie-Hellman Assumption (BDH)
[0096] Since the Decision Diffie-Hellman problem (DDH) in .sub.1 is
easy, embodiments of the invention do not use DDH to build
cryptosystems in the group .sub.1. Instead, the security in
embodiments of our IBE system is based on a novel variant of the
Computational Diffie-Hellman assumption called the Bilinear
Diffie-Hellman Assumption (BDH).
[0097] Bilinear Diffie-Hellman Problem. Let .sub.1, .sub.2 be two
groups of prime order q. Let : .sub.1.times..sub.1.fwdarw..sub.2 be
an admissible map and let P be a generator of .sub.1. The BDH
problem in (.sub.1, .sub.2, > is as follows: Given <P, aP,
bP, cP> for some a, b, c.epsilon.*.sub.q compute W= (P,
p).sup.abc.epsilon..sub.2. An algorithm A has advantage .epsilon.
in solving BDH in <.sub.1, .sub.2, if Pr[(P,aP,bP,cP)=
(P,P).sup.abc].gtoreq..epsilon. where the probability is over the
random choice of a, b, c in *.sub.q, the random choice of
P.epsilon.*.sub.1, and the random bits of . BDH Parameter
Generator. We say that a randomized algorithm is a BDH parameter
generator if (1) takes a security parameter k.epsilon..sup.+, (2)
runs in polynomial time in k, and (3) outputs a prime number q, the
description of two groups .sub.1, .sub.2 of order q, and the
description of an admissible map :
.sub.1.times..sub.1.fwdarw..sub.2. We denote the output of G by
(1.sup.k)=<q, .sub.1, .sub.2, >. The security parameter k is
used to determine the size of q; for example, one could take q to
be a random k-bit prime. For i=1, 2 we assume that the description
of the group .sub.i contains polynomial time (in k) algorithms for
computing the group action in .sub.i and contains a generator of
.sub.i. The generator of .sub.i enables us to generate uniformly
random elements in .sub.i. Similarly, we assume that the
description of contains a polynomial time algorithm for computing .
We give an example of a BDH parameter generator below in the
detailed example of an IBE system using the Weil pairing. Bilinear
Diffie-Hellman Assumption. Let be a BDH parameter generator. We say
that an algorithm has advantage .epsilon.(k) in solving the BDH
problem for if for sufficiently large k: Adv , .function. ( k ) =
Pr [ .function. ( q , 1 , 2 , e ^ , P , aP , bP , cP ) = e ^
.function. ( P , P ) abc | q , 1 , 2 , e ^ .rarw. .function. ( 1 k
) , P .rarw. 1 * , a , b , c .rarw. q * ] > .epsilon. .times.
.times. ( k ) ##EQU3## We say that satisfies the BDH assumption if
for any randomized polynomial time (in k) algorithm and for any
polynomial f.epsilon.[x] algorithm solves the BDH problem with
advantage at most 1/f(k). When satisfies the BDH assumption we say
that BDH is hard in groups generated by .
[0098] In the description below of a detailed example of an IBE
system we give some examples of BDH parameter generators that are
believed to satisfy the BDH assumption.
[0099] Hardness of BDH. It is interesting to study the relationship
of the BDH problem to other hard problems used in cryptography.
Currently, all we can say is that the BDH problem in <.sub.1,
.sub.2, > is no harder than the CDH problem in .sub.1 or .sub.2.
In other words, an algorithm for CDH in .sub.1 or .sub.2 is
sufficient for solving BDH in (.sub.1, .sub.2, ). The converse is
currently an open problem: is an algorithm for BDH sufficient for
solving CDH in .sub.1 or in .sub.2?
[0100] We note that in a detailed example of an IBE system below,
the isomorphisms from .sub.1 to .sub.2 induced by the admissible
map are believed to be one-way functions. More specifically, for a
point Q.epsilon.*.sub.1 define the isomorphism f.sub.Q:
.sub.1.fwdarw..sub.2 by f.sub.Q(P)= (P, Q). If any one of these
isomorphisms turns out to be invertible, then BDH is easy in
(.sub.1, .sub.2, >. Fortunately, an efficient algorithm for
inverting f.sub.Q would imply an efficient algorithm for deciding
DDH in the group .sub.2. In the exemplary embodiments DDH is
believed to be hard in the group .sub.2. Hence, the isomorphisms
f.sub.Q: .sub.1 .fwdarw..sub.2 induced by the admissible map are
believed to be one-way functions.
Exemplary Identity-Based Encryption Scheme
[0101] We describe the following exemplary embodiments in stages.
First we describe a basic identity-based encryption system and
method which is not secure against an adaptive chosen ciphertext
attack. Another embodiment described below extends the basic scheme
to get security against an adaptive chosen ciphertext attack
(IND-ID-CCA) in the random oracle model. We later relax some of the
requirements on the hash functions to provide alternative
embodiments. These embodiments are described with reference to a
generic BDH parameter generator satisfying the BDH assumption.
Later we describe a detailed example of an IBE system using the
Weil pairing.
BasicIdent
[0102] The following describes a basic embodiment, called
BasicIdent. We present the embodiment by describing the four
algorithms: Setup, Extract, Encrypt, Decrypt. We let k be the
security parameter given to the setup algorithm. We let be some BDH
parameter generator. [0103] Setup: Given a security parameter
k.epsilon..sup.+, the algorithm in the basic embodiment works as
follows: [0104] Step 1: Run on input k to generate a prime q, two
groups .sub.1, .sub.2 of order q, and an admissible map :
.sub.1.times..sub.1.fwdarw..sub.2. Choose an arbitrary generator
P.epsilon..sub.1. [0105] Step 2: Pick a random s.epsilon.*.sub.q
and set P.sub.pub=sP. [0106] Step 3: Choose a cryptographic hash
function H.sub.1: {0, 1}*.fwdarw.*.sub.1. Choose a cryptographic
hash function H.sub.2: .sub.2.fwdarw.{0, 1}.sup.n for some n. The
security analysis will view H.sub.1, H.sub.2 as random oracles.
[0107] The message space is ={0, 1}.sup.n. The ciphertext space is
=*.sub.1.times.{0, 1}.sup.n. The system parameters are
params=<q, .sub.1, .sub.2, , n, P, P.sub.pub, H.sub.1,
H.sub.2>. The master-key is s.epsilon.*.sub.q. Embodiments of
the IBE system may be used to encrypt a symmetric key, in which
case one may take n to be, for example, 128 or 256. For k one may
use, for example, 512 or 1024 or 2048. [0108] Extract: For a given
string ID.epsilon.{0, 1}* the algorithm in the basic embodiment
does: (1) computes Q.sub.ID=H.sub.1(ID).epsilon.*.sub.1, and (2)
sets the private key d.sub.ID to be d.sub.ID=sQ.sub.ID where s is
the master key. [0109] Extract may be performed by a PKG in various
embodiments as illustrated in FIG. 2. The PKG obtains the master
key in block 200, obtains the public identifier ID in block 210,
computes the public key from the ID in block 220, then computes the
private key from the master key and the public key in block 230. In
block 240 the private key is then sent to an entity associated with
the public identifier ID, normally after the entity's identity has
been authenticated. [0110] Encrypt: To encrypt M.epsilon. under the
public key ID do the following: (1) compute
Q.sub.ID=H.sub.1(ID).epsilon.*.sub.1, (2) choose a random
r.epsilon.*.sub.q, and (3) set the ciphertext to be
C=<rP,M.sym.H.sub.2(g.sub.ID.sup.r)> where g.sub.ID=
(Q.sub.ID,P.sub.pub).epsilon.*.sub.2. [0111] In the basic
embodiment, the sender of a message may perform Encrypt as
illustrated in FIG. 3. In block 300 the system parameters are
obtained (from an external resource such as a PKG, or from a local
storage medium if they were obtained previously). A receiver's ID
is determined in block 310, and the corresponding public key is
computed from the ID in block 320. Then the secret message key is
computed in block 330, and the message key is used to encrypt the
message in block 340. [0112] Decrypt: Let C=<U, V>.epsilon.C
be a ciphertext encrypted using the public key ID. To decrypt C
using the private key d.sub.ID.epsilon.*.sub.1 compute:
V.sym.H.sub.2( (d.sub.ID,U))=M. [0113] In the basic embodiment, the
receiver may perform Decrypt as illustrated in FIG. 4. In block
400, the system parameters are obtained (from an external resource
such as a PKG, or from a local storage medium if they were obtained
previously). The ciphertext V and an element rP are obtained from
the sender in block 410. The element rP may be considered a portion
of the total ciphertext obtained from the sender. In block 420 the
receiver obtains the private key d.sub.ID corresponding to the
public identifier ID used to encrypt the message. The private key
is normally obtained from an external resource such as a PKG, or
from a local storage medium if it was obtained previously. The
secret message key is then computed in block 430, and used to
decrypt the message in block 440. This completes the description of
BasicIdent for the basic embodiment. We first verify consistency.
When everything is computed as above we have: 1. During encryption
M is bitwise exclusive-ored with the hash of: g.sub.ID.sup.r. 2.
During decryption V is bitwise exclusive-ored with the hash of:
(d.sub.ID, U). These masks used during encryption and decryption
are the same since: (d.sub.ID,U)= (sQ.sub.ID,rP)=
(Q.sub.ID,p).sup.st= (Q.sub.ID,P.sub.pub).sup.r=g.sub.ID.sup.r
Thus, applying decryption after encryption produces the original
message M as required. Performance considerations of BasicIdent are
discussed later. [0114] Security. Next, we study the security of
this basic embodiment.
[0115] The security of the exemplary system is based on the
assumption that a variant of the Computational Diffie-Hellman
problem in .sub.1 is hard. The technical security details of the
encryption scheme are discussed by the inventors in D. Boneh, M.
Franklin, "Identity based encryption from the Weil pairing",
extended abstract in Advances in Cryptology--Crypto 2001, Lecture
Notes in Computer Science, Vol. 2139, Springer-Verlag, pp. 231-229,
2001, which is incorporated herein by reference.
[0116] In an exemplary embodiment, the performance of the system is
comparable to the performance of ElGamal encryption in *.sub.p. The
security of the exemplary system is based on a variant of the
computational Diffie-Hellman assumption. Based on this assumption
we show that the exemplary system has chosen ciphertext security in
the random oracle model. In accordance with a distributed PKG
embodiment, threshold cryptography techniques allow the PKG to be
distributed so that the master-key is never available in a single
location. Unlike common threshold systems, we show that robustness
for the distributed PKG embodiment is free.
[0117] To argue about the security of the exemplary system, we
define chosen ciphertext security for identity-based encryption.
Our model gives the adversary more power than the standard model
for chosen ciphertext security. First, we allow the attacker to
attack an arbitrary public key ID of her choice. Second, while
mounting a chosen ciphertext attack on ID we allow the attacker to
obtain from the PKG the private key for any public key of her
choice, other than the private key for ID. This models an attacker
who obtains a number of private keys corresponding to some
identities of her choice and then tries to attack some other public
key ID of her choice. Even with the help of such queries, it is
desirable that the attacker still have negligible advantage in
defeating the semantic security of the system.
[0118] The following theorem shows that BasicIdent is a
semantically secure identity based encryption scheme (IND-ID-CPA)
assuming BDH is hard in groups generated by . Theorem 1 Suppose the
hash functions H.sub.1, H.sub.2 are random oracles. Then BasicIdent
is a semantically secure identity based encryption scheme
(IND-ID-CPA) assuming BDH is hard in groups generated by .
Concretely, suppose there is an IND-ID-CPA adversary that has
advantage .epsilon.(k) against the scheme BasicIdent. Suppose makes
at most q.sub.E>0 private key extraction queries and
q.sub.H.sub.2>0 hash queries to H.sub.2. Then there is an
algorithm that solves BDH in groups generated by with advantage at
least: Adv , .function. ( k ) .gtoreq. 2 .times. .epsilon.
.function. ( k ) e .function. ( 1 + q E ) q H 2 ##EQU4## Here
e.apprxeq.2.71 is the base of the natural logarithm. The running
time of is O(time()).
[0119] To prove the theorem we first define a related Public Key
Encryption scheme (not an identity based scheme), called BasicPub.
BasicPub is described by three algorithms: keygen, encrypt,
decrypt. [0120] keygen: Given a security parameter
k.epsilon..sup.+, the algorithm works as follows: [0121] Step 1:
Run on input k to generate two prime order groups .sub.1, .sub.2
and an admissible map : .sub.1.times..sub.1.fwdarw..sub.2. Let q be
the order of .sub.1, .sub.2. Choose an arbitrary generator
P.epsilon..sub.1. [0122] Step 2: Pick a random s.epsilon.*.sub.q
and set P.sub.pub=sP. Pick a random Q.sub.ID.epsilon.*.sub.1.
[0123] Step 3: Choose a cryptographic hash function H.sub.2:
.sub.2{0, 1}.sup.n for some n. [0124] Step 4: The public key is
<q, .sub.1, .sub.2, , n, P, P.sub.pub, Q.sub.ID, H.sub.2>.
The private key is d.sub.ID=sQ.sub.ID.epsilon.*.sub.1. [0125]
encrypt: To encrypt M.epsilon.{0, 1}.sup.n choose a random
r.epsilon.*.sub.q and set the ciphertext to be:
C=<rP,M.sym.H.sub.2(g.sup.r)> where g=
(Q.sub.ID,P.sub.pub).epsilon.*.sub.2 [0126] decrypt: Let C=<U,
V> be a ciphertext created using the public key <q, .sub.1,
.sub.2, , n, P, P.sub.pub, Q.sub.ID, H.sub.2>. To decrypt C
using the private key d.sub.ID.epsilon.*.sub.1 compute:
V.sym.H.sub.2( (d.sub.ID,U))=M This completes the description of
BasicPub. We now prove Theorem 1 in two steps. We first show that
an IND-ID-CPA attack on BasicIdent can be converted to a IND-CPA
attack on BasicPub. This step shows that private key extraction
queries do not help the adversary. We then show that BasicPub is
IND-CPA secure if the BDH assumption holds. The proofs are omitted.
Lemma 2 Let H.sub.1 be a random oracle from {0, 1}* to *.sub.1. Let
be an IND-ID-CPA adversary that has advantage .epsilon.(k) against
BasicIdent. Suppose makes at most q.sub.E>0 private key
extraction queries. Then there is a IND-CPA adversary that has
advantage at least .epsilon.(k)/e(1+q.sub.E) against BasicPub. Its
running time is O(time()). Lemma 3 Let H.sub.2 be a random oracle
from .sub.2 to {0, 1}.sup.n. Let be an IND-CPA adversary that has
advantage .epsilon.(k) against BasicPub. Suppose makes a total of
q.sub.H.sub.2>0 queries to H.sub.2. Then there is an algorithm
that solves the BDH problem for with advantage at least
2.epsilon.(k)/q.sub.H.sub.2 and a running time O(time()). Proof of
Theorem 1. The theorem follows directly from Lemma 2 and Lemma 3.
Composing both reductions shows that an IND-ID-CPA adversary on
BasicIdent with advantage .epsilon.(k) gives a BDH algorithm for G
with advantage at least 2.epsilon.(k)/e(1+q.sub.E)q.sub.H.sub.2, as
required. Identity-Based Encryption with Chosen Ciphertext
Security
[0127] According to one embodiment of the invention, a technique of
Fujisaki and Okamoto (described in E. Fujisaki and T. Okamoto,
"Secure integration of asymmetric and symmetric encryption
schemes", in Advances in Cryptology--Crypto '99, Lecture Notes in
Computer Science, Vol. 1666, Springer-Verlag, pp. 537-554, 1999,
which is incorporated herein by reference) may be appropriately
adapted to convert the BasicIdent embodiment of the previous
section into a chosen ciphertext secure embodiment of an IBE system
(in the sense defined earlier) in the random oracle model. Let be a
probabilistic public key encryption scheme. We denote by .sub.pk(M;
r) the encryption of M using the random bits r under the public key
pk. Fujisaki-Okamoto define the hybrid scheme .sup.hy as:
.sub.pk.sup.hy(M)=<.sub.pk(.sigma.;H.sub.3(.sigma.,M)),H.sub.4(.sigma.-
).sym.M> Here .sigma. is generated at random and H.sub.3,
H.sub.4 are cryptographic hash functions. Fujisaki-Okamoto show
that if is a one-way encryption scheme then .sup.hy is a chosen
ciphertext secure system (IND-CCA) in the random oracle model
(assuming .sub.pk satisfies some natural constraints). We note that
semantic security implies one-way encryption and hence the
Fujisaki-Okamoto result also applies if is semantically secure
(IND-CPA).
[0128] We apply the Fujisaki-Okamoto transformation to BasicIdent
and show that the resulting embodiment of an IBE system is
IND-ID-CCA secure. We obtain the following IBE embodiment which we
call FullIdent. Recall that n is the length of the message to be
encrypted. [0129] Setup: As in the BasicIdent scheme. In addition,
we pick a hash function H.sub.3: {0, 1}.sup.n.times.{0,
1}.sup.n.fwdarw.*.sub.q, and a hash function H.sub.4: {0,
1}.sup.n.fwdarw.{0, 1}.sup.n. [0130] Extract: As in the BasicIdent
scheme. [0131] Encrypt: To encrypt M.epsilon.{, 1}.sup.n under the
public key ID do the following: (1) compute
Q.sub.ID=H.sub.1(ID).epsilon.*.sub.1, (2) choose a random
.sigma..epsilon.{0, 1}.sup.n, (3) set r=H.sub.3(.sigma., M), and
(4) set the ciphertext to be
C=<rP,.sigma..sym.H.sub.2(g.sub.ID.sup.r),
M.sym.(H.sub.4(.sigma.)> where g.sub.ID=
(Q.sub.ID,P.sub.pub).epsilon..sub.2 [0132] Decrypt: Let C=<U, V,
W> be a ciphertext encrypted using the public key ID. If
U.noteq.*.sub.1 reject the ciphertext. To decrypt C using the
private key d.sub.ID.epsilon.*.sub.1 do: [0133] 1. Compute
V.sym.H.sub.2( (d.sub.ID, U))=.sigma.. [0134] 2. Compute
W.sym.H.sub.4(.sigma.)=M. [0135] 3. Set r=H.sub.3(.sigma., M). Test
that U=rP. If not, reject the ciphertext. [0136] 4. Output M as the
decryption of C. This completes the description of FullIdent. Its
implementation follows the same basic pattern as BasicIdent shown
in FIGS. 2, 3, 4. Note that M is encrypted as
W=M.sym.H.sub.4(.sigma.). This can be replaced by
W=E.sub.H.sub.4.sub.(.sigma.)(M) where E is a semantically secure
symmetric encryption scheme. [0137] Security. The following theorem
shows that FullIdent is a chosen ciphertext secure IBE (i.e.
IND-ID-CCA), assuming BDH is hard in groups generated by . Theorem
4 Let the hash functions H.sub.1, H.sub.2, H.sub.3, H.sub.4 be
random oracles. Then FullIdent is a chosen ciphertext secure IBE
(IND-ID-CCA) assuming BDH is hard in groups generated by .
Concretely, suppose there is an IND-ID-CCA adversary that has
advantage .epsilon.(k) against the scheme FullIdent and runs in
time at most t(k). Suppose makes at most q.sub.E extraction
queries, at most q.sub.D decryption queries, and at most
q.sub.H.sub.2, q.sub.H.sub.3, q.sub.H.sub.4 queries to the hash
functions H.sub.2, H.sub.3, H.sub.4 respectively. Then there is a
BDH algorithm for with running time t.sub.1(k) where: Adv ,
.function. ( k ) .gtoreq. 2 .times. .times. FO adv .function. (
.epsilon. .function. ( k ) e .function. ( 1 + q E + q D ) , q H 4 ,
q H 3 , q D ) / q H 2 ##EQU5## t 1 .function. ( k ) .ltoreq. FO
time .function. ( t .function. ( k ) , q H 4 , q H 3 ) ##EQU5.2##
where the functions FO.sub.time and FO.sub.adv are defined in
Theorem 5.
[0138] The proof of Theorem 4 is based on the following result of
Fujisaki and Okamoto. Let BasicPub.sup.hy be the result of applying
the Fujisaki-Okamoto transformation to BasicPub. Theorem 5
(Fujisaki-Okamoto) Suppose is an IND-CCA adversary that achieves
advantage .epsilon.(k) when attacking BasicPub.sup.hy. Suppose has
running time t(k), makes at most q.sub.D decryption queries, and
makes at most q.sub.H.sub.3, q.sub.H.sub.4 queries to the hash
functions H.sub.3, H.sub.4 respectively. Then there is an IND-CPA
adversary against BasicPub with running time t.sub.1(k) and
advantage .epsilon..sub.1(k) where .epsilon. 1 .function. ( k )
.gtoreq. .times. FO adv .function. ( .epsilon. .function. ( k ) , q
H 4 , q H 3 , q D ) = .times. 1 2 .times. ( q H 4 + q H 3 )
.function. [ ( .epsilon. .function. ( k ) + 1 ) .times. ( 1 - 2 / q
) q D - 1 ] ##EQU6## t 1 .function. ( k ) .ltoreq. .times. FO time
.function. ( t .function. ( k ) , q H 4 , q H 3 ) = .times. t
.function. ( k ) + O .function. ( ( q H 4 + q H 3 ) n ) ,
##EQU6.2## Here q is the size of the groups .sub.1, .sub.2 and n is
the length of .sigma..
[0139] In fact, Fujisaki-Okamoto prove a stronger result: Under the
hypothesis of Theorem 5, BasicPub.sup.hy would not even be a
one-way encryption scheme. For our purposes the result in Theorem 5
is sufficient. To prove Theorem 4 we also need the following lemma
to translate between an IND-ID-CCA chosen ciphertext attack on
FullIdent and an IND-CCA chosen ciphertext attack on
BasicPub.sup.hy. Lemma 6 Let A be an IND-ID-CCA adversary that has
advantage .epsilon.(k) against FullIdent. Suppose makes at most
q.sub.E>0 private key extraction queries and at most q.sub.D
decryption queries. Then there is an IND-CCA adversary that has
advantage at least .epsilon. .function. ( k ) e .function. ( 1 + q
E + q D ) ##EQU7## against BasicPub.sup.hy. Its running time is
O(time()). Proof of Theorem 4. By Lemma 6 an IND-ID-CCA adversary
on FullIdent implies an IND-CCA adversary on BasicPub.sup.hy. By
Theorem 5 an IND-CCA adversary on BasicPub.sup.hy implies an
IND-CPA adversary on BasicPub. By Lemma 3 an IND-CPA adversary on
BasicPub implies an algorithm for BDH. Composing all these
reductions gives the required bounds. .quadrature.
[0140] Relaxing the Hashing Requirements
[0141] Recall that an IBE system of Section uses a hash function
H.sub.1: {0, 1}*.fwdarw.*.sub.1. The detailed example of an IBE
system presented in the next section uses .sub.1 as a subgroup of
the group of points on an elliptic curve. In practice, it sometimes
can be difficult to build hash functions that hash directly onto
such groups. In an exemplary embodiment, we therefore show how to
relax the requirement of hashing directly onto *.sub.1. Rather than
hash onto *.sub.1 we hash onto some set A.OR right.{0, 1}* and then
use a deterministic encoding function to map A onto *.sub.1.
Admissible encodings: Let .sub.1 be a group and let A.epsilon.{0,
1}* be a finite set. We say that an encoding function L:
A.fwdarw.*.sub.1 is admissible if it satisfies the following
properties:
[0142] 1. Computable: There is an efficient deterministic algorithm
to compute L(z) for any x.epsilon.A. [0143] 2. l-to-1: For any
y.epsilon.*.sub.1 the preimage of y under L has size exactly l. In
other words, |L.sup.-1(y)|=l for all y.epsilon.*.sub.1. Note that
this implies that |A|=l|*.sub.1|. [0144] 3. Samplable: There is an
efficient randomized algorithm .sub.s such that .sub.s(y) induces a
uniform distribution on L.sup.-1(y) for any y.epsilon.*.sub.1. In
other words, .sub.s(y) is a uniform random element in L.sup.-1(y).
We modify FullIdent to obtain an IND-ID-CCA secure embodiment of an
IBE system where H.sub.1 is replaced by a hash function into some
set A. Since the change is relatively minor we refer to this new
scheme as FullIdent: [0145] Setup: As in the FullIdent embodiment.
The only difference is that H.sub.1 is replaced by a hash function
H'.sub.1: {0, 1}*.fwdarw.A. The system parameters also include a
description of an admissible encoding function L: A.fwdarw.*.sub.1.
[0146] Extract, Encrypt: As in the FullIdent embodiment. The only
difference is that in Step 1 these algorithms compute
Q.sub.ID=L(H'.sub.1(ID)).epsilon.*.sub.1. [0147] Decrypt: As in the
FullIdent embodiment. This completes the description of FullIdent'.
The following theorem shows that FullIdent' is a chosen ciphertext
secure IBE (i.e. IND-ID-CCA), assuming FullIdent is. Theorem 7 Let
be an IND-ID-CCA adversary on FullIdent' that achieves advantage
.epsilon.(k). Suppose makes at most q.sub.H.sub.1 queries to the
hash function H'.sub.1. Then there is an IND-ID-CCA adversary on
FullIdent that achieves the same advantage .epsilon.(k) and
time()=time()+q.sub.H.sub.1time(L.sub.s) Proof Sketch Algorithm
attacks FullIdent by running algorithm . It relays all decryption
queries, extraction queries, and hash queries from directly to the
challenger and relays the challenger's response back to . It only
behaves differently when issues a hash query to H'.sub.1. Recall
that only has access to a hash function H.sub.1: {0,
1}*.fwdarw.*.sub.1. To respond to H'.sub.1 queries algorithm
maintains a list of tuples <ID.sub.j, y.sub.j> as explained
below. We refer to this list as the (H'.sub.1).sup.list. The list
is initially empty. When queries the oracle H'.sub.1 at a point
ID.sub.i algorithm responds as follows: [0148] 1. If the query
ID.sub.i already appears on the (H'.sub.1).sup.list in a tuple
<ID.sub.i, y.sub.i> then respond with
H'.sub.1(ID.sub.i)=y.sub.i.epsilon.A. [0149] 2. Otherwise, issues a
query for H.sub.1(ID.sub.i). Say,
H.sub.1(ID.sub.i)=.alpha..epsilon.*.sub.1. [0150] 3. runs the
sampling algorithm .sub.s(.alpha.) to generate a random element
y.epsilon.L.sup.-1(.alpha.). [0151] 4. adds the tuple <ID.sub.i,
y> to the (H'.sub.1).sup.list and responds to with
H'.sub.1(ID.sub.i)=y.epsilon.A. Note that y is uniformly
distributed in A as required since .alpha. is uniformly distributed
in *.sub.1 and L is an l-to-1 map. Algorithm 's responses to all of
's queries, including H'.sub.1 queries, are identical to 's view in
the real attack. Hence, will have the same advantage .epsilon.(k)
in winning the game with the challenger. A Detailed Example of an
IBE System Using the Weil Pairing
[0152] In this section we use FullIdent' to describe a detailed
example of an embodiment of an IBE system. This embodiment is based
on the Weil pairing. Although in practice the Tate pairing has
computational advantages and may be used instead of the Weil
pairing in various embodiments, the implementation using the Weil
pairing will be described first because it is simpler. Later, the
Tate pairing will be discussed.
Properties of the Weil Pairing
[0153] Let p>3 be a prime satisfying p=2 mod 3 and let q be some
prime factor of p+1. Let E be the elliptic curve defined by the
equation y.sup.2=x.sup.3+1 over .sub.p. We state a few elementary
facts about this curve E. From here on we let E(.sub.p.sub.r)
denote the group of points on E defined over .sub.p.sub.r. [0154]
Fact 1: Since x.sup.3+1 is a permutation on .sub.p it follows that
the group E(.sub.p) contains p+1 points. We let O denote the point
at infinity. Let P.epsilon.E(.sub.p) be a point of order q and let
.sub.1 be the subgroup of points generated by P. [0155] Fact 2: For
any y.sub.0.epsilon..sub.p there is a unique point (x.sub.0,
y.sub.0) on E(.sub.p), namely
x.sub.0=(y.sub.0.sup.2-1).sup.1/3.epsilon..sub.p. Hence, if (x, y)
is a random non-zero point on E(.sub.p) then y is uniform in
.sub.p. We use this property to build a simple admissible encoding
function. [0156] Fact 3: Let 1.noteq..zeta..epsilon..sub.p.sub.2 be
a solution of x.sup.3-1=0 mod p. Then the map .phi.(x, y)=(.zeta.x,
y) is an automorphism of the group of points on the curve E. Note
that for any point Q=(x, y).epsilon.E(.sub.p) we have that
.phi.(Q).epsilon.E(.sub.p.sub.2), but .phi.(Q).epsilon.E(.sub.p).
Hence, Q.epsilon.E(.sub.p) is linearly independent of
.phi.(Q).epsilon.E.sub.p.sub.2). [0157] Fact 4: Since the points
P.epsilon..sub.1 and .phi.(P) are linearly independent they
generate a group isomorphic to .sub.q.times..sub.q. We denote this
group of points by E[q]. Let .sub.2 be the subgroup of
*.sub.p.sub.2 of order q. The Weil pairing on the curve
E(.sub.p.sub.2) is a mapping e: E[q].times.E[q].fwdarw..sub.2.
(This map is defined and discussed in the section below entitled
Description of the Weil Pairing.) For any Q, R.epsilon.E(.sub.p)
the Weil pairing satisfies e(Q, R)=1. In other words, the Weil
pairing is degenerate on E(.sub.p), and hence degenerate on the
group .sub.1. To get a non-degenerate map we define the modified
Weil pairing : .sub.1.times..sub.1.fwdarw..sub.2 as follows:
(P,Q)=e(P,.phi.(Q)) The modified Weil pairing satisfies the
following properties: [0158] 1. Bilinear: For all P,
Q.epsilon..sub.1 and for all a, b.epsilon. we have (aP, bQ)= (P,
Q).sup.ab. [0159] 2. Non-degenerate: If P is a generator of .sub.1
then (P, P).epsilon.*.sub.p.sub.2 is a generator of .sub.2. [0160]
3. Computable: Given P, Q.epsilon..sub.1 there is an efficient
algorithm to compute (P, Q).epsilon..sub.2. (This algorithm is
described in the section below entitled Description of the Weil
Pairing.) Its running time is comparable to exponentiation in
.sub.p. Although the Computational Diffie-Hellman problem (CDH)
appears to be hard in the group .sub.1, the Decision Diffie-Hellman
problem (DDH) is easy in .sub.1. BDH Parameter Generator .sub.1:
Given a security parameter 2<k.epsilon.the BDH parameter
generator picks a random k-bit prime q and finds the smallest prime
p such that (1) p=2 mod 3, (2) q divides p+1, and (3) q.sup.2 does
not divide p+1. We write p=lq+1. The group .sub.1 is the subgroup
of order q of the group of points on the curve y.sup.2=x.sup.3+1
over .sub.p. The group .sub.2 is the subgroup of order q of
*.sub.p.sub.2. The bilinear map : .sub.1.times..sub.1.fwdarw..sub.2
is the modified Weil pairing defined above.
[0161] The BDH parameter generator .sub.1 is believed to satisfy
the BDH assumption asymptotically. However, there is still the
question of what values of p and q can be used in practice to make
the BDH problem sufficiently hard. It is desirable that we can
ensure, at the very least, that the discrete log problem in .sub.1
is sufficiently hard. As pointed out earlier, the discrete log
problem in .sub.1 is efficiently reducible to discrete log in
.sub.2. Hence, computing discrete log in *.sub.p.sub.2 is
sufficient for computing discrete log in .sub.1. In practice, for
proper security of discrete log in *.sub.p.sub.2 it is desirable to
use primes p that are at least 512-bits long (so that the group
size is at least 1024 bits long). Consequently, in some
embodiments, this BDH parameter generator is used with primes p
that may be 512-bits long or more.
An Admissible Encoding Function: MapToPoint
[0162] Let .sub.1, .sub.2 be two groups generated by .sub.1 as
defined above. Recall that an IBE system discussed earlier uses a
hash function H.sub.1: {0, 1}*.fwdarw.*.sub.1. It suffices to have
a hash function H.sub.1: {0, 1}*.fwdarw.A for some set A, and an
admissible encoding function L: A.fwdarw.*.sub.1. In what follows
the set A will be .sub.p, and the admissible encoding function L
will be called MapToPoint, which may be used in various embodiments
of the present invention.
[0163] In this example, let p be a prime satisfying p=2 mod 3 and
p=lq-1 for some prime q>3. In this exemplary embodiment, q does
not divide l (i.e. q.sup.2 does not divide p+1). Let E be the
elliptic curve y.sup.2=x.sup.3+1 over .sub.p. Let .sub.1 be the
subgroup of points on E of order q. In addition, a hash function
H.sub.1: {0, 1}*.fwdarw..sub.p is provided.
In this exemplary embodiment, algorithm MapToPoint works as follows
on input y.sub.0.epsilon..sub.p:
1. Compute
x.sub.0=(y.sub.0.sup.2-1).sup.1/3=(y.sub.0.sup.2-1).sup.(2p-1)/3.epsilon.-
.sub.p.
2. Let Q=(x.sub.0, y.sub.0).epsilon.E(.sub.p) and set
Q.sub.ID=lQ.epsilon..sub.1.
3. Output MapToPoint (y.sub.0)=Q.sub.ID.
This completes the description of MapToPoint.
[0164] We note that there are l-1 values of y.sub.0.epsilon..sub.p
for which lQ=l(x.sub.0, y.sub.0)=O (these are the non-O points of
order dividing l). Let B.OR right..sub.p be the set of these
y.sub.0. When H.sub.1(ID) is one of these l-1 values Q.sub.ID is
the identity element of .sub.1. It is extremely unlikely for
H.sub.1(ID) to hit one of these points--the probability is 1/{dot
over (q)}<1/2.sup.k. Hence, for simplicity we say that
H.sub.1(ID) only outputs elements in .sub.p\B, i.e. H.sub.1: {0,
1}*.fwdarw..sub.p\B. In other embodiments, algorithm MapToPoint can
be easily extended to handle the values y.sub.0.epsilon.B by
hashing ID multiple times using different hash functions.
Proposition 8 MapToPoint: .sub.p\B.fwdarw.*.sub.1 is an admissible
encoding function.
[0165] Proof The map is clearly computable and is a l-to-1 mapping.
It remains to show that L is samplable. Let P be a generator of
E(.sub.p). Given a Q.epsilon.*.sub.1 the sampling algorithm .sub.s
does the following: (1) pick a random b.epsilon.{0, . . . , l-1},
(2) compute Q'=l.sup.-1Q+bqP=(x, y), and (3) output
.sub.s(Q)=y.epsilon..sub.p. Here l.sup.-1 is the inverse of l in
*.sub.q. This algorithm outputs a random element from the l
elements in MapToPoint.sup.-1(Q) as required. .quadrature.
A Detailed Example of an IBE System
[0166] Using the BDH parameter generator .sub.1 and the admissible
encoding function MapToPoint we obtain the following detailed
example of an embodiment of an IBE system. [0167] Setup: Given a
security parameter k.epsilon..sup.+, the algorithm works as
follows: [0168] Step 1: Run .sub.1 on input k to generate a k-bit
prime q and a prime p=2 mod 3 such that q divides p+1. Let E be the
elliptic curve defined by y.sup.2=x.sup.3+1 over .sub.p. Choose an
arbitrary P.epsilon.E(.sub.p) of order q. [0169] Step 2: Pick a
random s.epsilon.*.sub.q and set P.sub.pub=sP. [0170] Step 3: Pick
four hash functions: H.sub.1: {0, 1}*.fwdarw..sub.p; H.sub.2:
.sub.p.sub.2.fwdarw.{0, 1}.sup.n for some n H.sub.3: {0,
1}.sup.n.times.{0, 1}.sup.n.fwdarw.*.sub.q, and a hash function
H.sub.4: {0, 1}.sup.n.fwdarw.{0, 1}.sup.n. [0171] The message space
is ={0, 1}.sup.n. The ciphertext space is =E(.sub.p).times.{0,
1}.sup.n. The system parameters are params=<p, q, n, P,
P.sub.pub, H.sub.1, . . . , H.sub.4>. The master-key is
s.epsilon.*.sub.q. [0172] Extract: For a given string
ID.epsilon.{0, 1}* the algorithm builds a private key d as follows:
[0173] Step 1: Compute
MapToPoint(H.sub.1(ID))=Q.sub.ID.epsilon.E(.sub.p) of order q.
[0174] Step 2: Set the private key d.sub.ID to be
d.sub.ID=sQ.sub.ID where s is the master key. [0175] Encrypt: To
encrypt M.epsilon.{0, 1}.sup.n under the public key ID do the
following: [0176] Step 1: Compute
MapToPoint(H.sub.1(ID))=Q.sub.ID.epsilon.E(.sub.p) of order q.
[0177] Step 2: Choose a random .sigma..epsilon.{0, 1}.sup.n. [0178]
Step 3: Set r=H.sub.3(.sigma., M). [0179] Step 4: Set the
ciphertext to be
C=<rP,.sigma..sym.H.sub.2(g.sub.ID.sup.r),M.sym.H.sub.4(.sigma.)>
where g.sub.ID= (Q.sub.ID,P.sub.pub).epsilon..sub.p.sub.2 [0180]
Decrypt: Let C=<U, V, W>.epsilon. be a ciphertext encrypted
using the public key ID. If U.epsilon.E(.sub.p) is not a point of
order q reject the ciphertext. To decrypt C using the private key
d.sub.ID do: Step 1. Compute V.sym.H.sub.2( (d.sub.ID, U))=.sigma..
Step 2. Compute W.sym.(H.sub.4(.sigma.)=M. Step 3. Set
r=H.sub.3(.sigma., M). Test that U=rP. If not, reject the
ciphertext. Step 4. Output M as the decryption of C. Performance.
In this embodiment, algorithms Setup and Extract are very simple.
At the heart of both algorithms is a standard multiplication on the
curve E(.sub.p). Algorithm Encrypt requires that the encryptor
compute the Weil pairing of Q.sub.ID and P.sub.pub. Note that this
computation is independent of the message, and hence can be done
once and for all. Once g.sub.ID is computed the performance of this
embodiment is almost identical to standard ElGamal encryption. We
also note that the ciphertext length of the exemplary embodiment of
BasicIdent is the same as in regular ElGamal encryption in .sub.p.
Decryption is a simple Weil pairing computation. Security. The
security of the detailed exemplary embodiment just described
follows directly from Theorem 4 and Theorem 7. Corollary 9 The
detailed exemplary embodiment described above is a chosen
ciphertext secure IBE system (i.e. IND-ID-CCA in the random oracle
model) assuming the BDH parameter generator .sub.1 satisfies the
BDH assumption. Extensions and Observations Tate Pairing and Other
Curves.
[0181] Embodiments of our IBE system work with efficiently
computable bilinear maps : .sub.1.times..sub.1.fwdarw..sub.2
between two groups .sub.1, .sub.2 where the BDH assumption holds.
Many different elliptic curves may give rise to such maps. For
example, one could use the curve y.sup.2=x.sup.3+x over .sub.p with
p=3 mod 4 and its endomorphism .phi.: (x, y).fwdarw.(-x, iy) where
i.sup.2=-1.
[0182] In an alternative embodiment, one may use a family of
nonsupersingular elliptic curves over .sub.p discussed by Miyaji et
al. (A. Miyaji, M. Nakabayashi, S. Takano, "New explicit condition
of elliptic curve trace for FR-reduction", IEICE Trans.
Fundamentals, Vol. E84 A, No. 5, May 2001). For example, to use a
curve E/.sub.p in this family one can take .sub.1 to be a cyclic
subgroup of E(.sub.p.sub.6) (that is not contained in E(.sub.p))
and then use the trace map on the curve E as the endomorphism .phi.
used to define the pairing . We also note that both encryption and
decryption in FullIdent can be made faster in alternate embodiments
by using the Tate pairing on elliptic curves rather than the Weil
pairing. In other embodiments, suitable bilinear maps may be
derived from abelian varieties.
Asymmetric Pairings
[0183] As mentioned earlier, embodiments of our IBE system are not
limited to symmetric maps, but may include asymmetric maps as well.
In other words, embodiments generally may use maps of the form :
.sub.0.times..sub.1.fwdarw..sub.2 where .sub.0, .sub.1 are groups
of prime order q. When .sub.0 and are equal we say the map is
symmetric. When .sub.0 and .sub.1 are not equal we say the map is
asymmetric.
[0184] The elements Q.sub.ID and P in the asymmetric case are
members of .sub.0 and .sub.1, respectively (or vice versa), and the
target group of the hash function H.sub.1 is selected accordingly.
However, to make the proof of security go through (Lemma 2 in
particular) we use a slightly strange looking complexity assumption
which we call the co-BDH assumption: given random P, aP,
bP.epsilon..sub.1 and Q, aQ, cQ.epsilon..sub.0 no polynomial time
algorithm can compute (P, Q).sup.abc with non-negligible
probability. If one is uses this assumption then for embodiments
using a curve E/.sub.p from Miyaji et al. (as just described above)
one can take .sub.1 to be a cyclic subgroup of E(.sub.p) of order q
and .sub.0 to be a different cyclic subgroup of E(.sub.p.sub.6) of
order q. This will result in a more efficient system than the
method described in the preceding paragraph for using these
curves.
Distributed PKG
[0185] In exemplary embodiments of an IBE system it is desirable
that the master-key stored at the PKG be protected. One way of
protecting this key is by distributing it among different sites
using techniques of threshold cryptography. Embodiments of our IBE
system support this in a very efficient and robust way. Recall that
in some embodiments discussed above, the master-key may be some
s.epsilon.*.sub.q and the PKG uses the group action to compute a
private key from s and Q.sub.ID, where Q.sub.ID is derived from the
user's public key ID. For example, d.sub.ID=sQ.sub.ID. A
distributed PKG embodiment can be implemented in a t-out-of-n
fashion by giving each of the n PKGs one share s.sub.i of a Shamir
secret sharing of s mod q. Each of the n PKGs can use its share
s.sub.i of the master key to generate a corresponding share d.sub.i
of a private key d.sub.ID by calculating d.sub.i=s.sub.iQ.sub.ID.
The user can then construct the entire private key d.sub.ID by
requesting from t of the n PKGs its share d.sub.i of the private
key, then combining the shares by calculating d ID = i .times.
.lamda. i .times. d i , ##EQU8## where the .lamda..sub.i's are the
appropriate Lagrange interpolation coefficients.
[0186] Furthermore, it is easy to make this embodiment robust
against dishonest PKGs using the fact that DDH is easy in .sub.1.
During setup each of the n PKGs publishes P.sub.i=s.sub.iP. During
a key generation request the user can verify that the response from
the i'th PKG is valid by testing that: (d.sub.i,P)=
(Q.sub.ID,P.sub.i) Thus, a misbehaving PKG will be immediately
caught. There is no need for zero-knowledge proofs as in regular
robust threshold schemes. The PKG's master-key can be generated in
a distributed fashion using the techniques of R. Gennaro et al. (R.
Gennaro, S. Jarecki, H. Krawczyk, T. Rabin, "Secure Distributed Key
Generation for Discrete-Log Based Cryptosystems", Advances in
Cryptology--Eurocrypt '99, Lecture Notes in Computer Science, Vol.
1592, Springer-Verlag, pp. 295-310, 1999). Using this technique,
the PKGs can execute a cooperative protocol that allows them to
jointly generate their respective shares of the master key without
the master key ever existing in any one place.
[0187] Note that a distributed master-key embodiment also enables
threshold decryption on a per-message basis, without any need to
derive the corresponding decryption key. For example, threshold
decryption of BasicIdent ciphertext (U, V) is straightforward if
each PKG responds with (s.sub.iQ.sub.ID, U).
[0188] FIG. 5 is a block diagram illustrating a distributed PKG
system, according to an embodiment of the invention. FIG. 5
includes a sender system 501, receiver system 502 and three PKGs
(PKG A 503, PKG B 504 and PKG C 505). In one embodiment
illustrating a 2-out-of-3 sharing, each of three PKGs contains a
different share of a master key, and any two of the three are able
to derive the master key. As shown in the figure, PKG A 503, PKG B
504, and PKG C 505 include, respectively, master key share s.sub.1,
511, master key share 52, 512, and master key share s.sub.3, 513.
In 2-out-of-3 sharing, any two out of these three PKGs could
combine their shares to determine the master key, although in this
embodiment each PKG secretly holds its master key share.
[0189] Sender system 501 sends a message to receiver 502. The
message 514 may be encrypted using a public key based on an
identifier ID of the receiver. In order to obtain the corresponding
private key, the receiver system queries two of the three PKGs
using, for example, the receiver's identity or public key. As shown
in the figure, receiver system 502 makes queries 506 and 507 to PKG
A 503 and PKG B 504, respectively, in order to obtain two shares of
the private key. In response to the queries, PKG A 503 and PKG B
504 return, respectively, share d.sub.1, 508, and share d.sub.2,
509, of private key d, 510. Receiver system 502 is then able to
assemble the corresponding private key d.sub.ID, which corresponds
to the public key with which the message 514 was encrypted. More
generally, the receiver could have selected to query any two of the
three PKGs. For example, receiver system 502 alternatively could
have queried PKGs B and C and combined private key shares d.sub.2
and d.sub.3 to produce the private key 510. These techniques easily
generalize to provide similar embodiments using t-out-of n
sharing.
[0190] Sender system 501, receiver system 502 as well as PKGs 503,
504 and 505 may be each implemented as computer systems which
include elements such as processors and computer-readable media
such as memory and other storage devices. Communication between the
respective elements may take place using data packets sent over
data networks, or any of various other forms of electronic and data
transmission and communication. The communication may transpire
over various architectures of communication, such as a computer
network, such as the Internet, with various wired, wireless and
other communications media.
Working in Subgroups
[0191] In an alternative embodiment of the detailed IBE system
described above, performance may be improved by working in a
comparatively small subgroup of the curve. For example, choose a
1024-bit prime p=2 mod 3 with p=aq-1 for some 160-bit prime q. The
point P is then chosen to be a point of order q. Each public key ID
is converted to a group point by hashing ID to a point Q on the
curve and then multiplying the point by a. The system is secure if
the BDH assumption holds in the group generated by P. The advantage
of this embodiment is that the Weil computation is done on points
of small order, and hence is much faster.
IBE Implies Signatures
[0192] Various IBE techniques described above can be used to
provide public key signature systems and methods. The intuition is
as follows. The private key for the signature scheme is the master
key for the IBE scheme. The public key for the signature scheme is
the set of global system parameters for the IBE scheme. The
signature on a message M is the IBE decryption key for ID=M. To
verify a signature, choose a random message M', encrypt M' using
the public key ID=M, and then attempt to decrypt using the given
signature on M as the decryption key. If the IBE system is
IND-ID-CCA, then the signature scheme is existentially unforgeable
against a chosen message attack. Note that, unlike most signature
schemes, this signature verification embodiment is randomized. This
shows that the IBE techniques described herein may encompass both
public key encryption and digital signatures. Signature schemes
derived from these approaches can be used to provide interesting
properties, as described by Boneh et al. (D. Boneh, B. Lynn, H.
Shacham, "Short signatures from the Weil pairing", in Advances in
Cryptology--AsiaCrypt 2001, Lecture Notes in Computer Science, Vol.
2248, Springer-Verlag, pp. 514-532, 2001, which is incorporated
herein by reference).
Escrow ElGamal Encryption
[0193] In this section we show that various IBE techniques
described above can be used to provide an ElGamal encryption system
embodiment having global escrow capability. In this embodiment, a
single escrow key enables the decryption of ciphertexts encrypted
under any public key.
[0194] In one exemplary embodiment, the ElGamal escrow system works
as follows. The Setup is similar to that for BasicIdent. Unlike the
identity-based BasicIdent, each user selects a secret random number
and uses it to generate a public/private key pair. A sender and
receiver can then use Encrypt and Decrypt to communicate an
encrypted message. The message is secure except for an escrow who
can use a master key s to decrypt the message.
[0195] FIG. 6 is a block diagram illustrating elements in a
cryptosystem with escrow decryption capability according to an
embodiment of the invention. The system includes a sender system
601 with encryption logic 610, receiver system 602 with decryption
logic 611, escrow agent system 604 and broadcasting system 605.
Broadcast system 605 sends system parameters to participants such
as escrow agent system 604, receiver system 602 and sender system
601. The receiver system 602 selects a private key x, 607, and uses
it to generate a public key P.sub.pub=xP, 606, which is then
published. The private key x and the public key P.sub.pub form a
complementary key pair. Using the public key P.sub.pub=xP, 606,
sender system 601 encrypts a message M with encryption logic 610.
Sender system 601 sends a resulting encrypted message 603 to
receiver 602. Receiver system 602 decrypts the message with
decryption logic 611 using the private key x, 607. Escrow agent
system 604 may intercept message 603 and, using the escrow agent
key s, 609, public key P.sub.pub=xP, 606, and decrypt message 603
with decryption logic 612. In an alternate embodiment, broadcast
system 605 and escrow agent 604 may be a single entity. In yet
another embodiment, the escrow agent key s may be shared in a
manner such as in the distributed PKG embodiments described
earlier.
[0196] In more detail, an exemplary embodiment of the technique
involves the following procedures: [0197] Setup: Let be some BDH
parameter generator. Given a security parameter k.epsilon..sup.+,
the algorithm works as follows: [0198] Step 1: Run on input k to
generate a prime q, two groups .sub.1, .sub.2 of order q, and an
admissible map : .sub.1.times..sub.1.fwdarw..sub.2. Let P be some
generator of .sub.1. [0199] Step 2: Pick a random s.epsilon.*.sub.q
and set Q=sP. [0200] Step 3: Choose a cryptographic hash function
H: .sub.2.fwdarw.{0, 1}.sup.n. [0201] The message space is ={0,
1}.sup.n. The ciphertext space is =.sub.1.times.{0, 1}.sup.n. The
system parameters are params=<q, .sub.1, .sub.2, , n, P, Q,
H>. The escrow key is s.epsilon.*.sub.q. [0202] keygen: A user
generates a public/private key pair for herself by picking a random
x.epsilon.*.sub.q and computing P.sub.pub=xP.epsilon..sub.1. Her
private key is x (or xQ), her public key is P.sub.pub. [0203]
Encrypt: To encrypt M.epsilon.{0, 1}.sup.n under the public key
P.sub.pub do the following: (1) pick a random r.epsilon.*.sub.q,
and (2) set the ciphertext to be: C=<rP,M.sym.H(g.sup.r)>
where g= (P.sub.pub,Q).epsilon..sub.2. [0204] This encryption
technique is also illustrated in FIG. 7, where the sender obtains
the system parameters and elements P and Q=sP in block 700, and
obtains the recipient's public key P.sub.pub=xP in block 710. The
sender then selects a random r and computes a message key in block
720. The message key is then used to encrypt a message in block
730. The sender then transmits an encapsulated key rP and encrypted
message V to the receiver. [0205] Decrypt: Let C=<U, V> be a
ciphertext encrypted using P.sub.pub. Then U.epsilon..sub.l. To
decrypt C using the private key x do: V.sym.H( (U,xQ))=M. [0206] As
illustrated in FIG. 8, the receiver obtains the system parameters
and elements P and Q=sP in block 800, then obtains the encrypted
message V and encapsulated key rP from the sender in block 810. The
receiver then computes the message key in block 820, and uses it to
decrypt the message in block 830. [0207] To see that the message
keys computed by the sender and receiver are the same, note that
the sender knows the secret r as well as the public Q=sP and
P.sub.pub=xP, and uses these to compute a key from (sP, xP).sup.r.
The receiver, on the other hand, knows the secret x as well as the
public Q=sP and rP, and uses these to compute a message key from
(rP, x(sP)). The bilinearity of implies that (sP, xP).sup.r= (rP,
x(sP)), so the sender and receiver compute the same message key.
[0208] Escrow-decrypt: The purpose of this embodiment is to permit
escrow decryption of otherwise secure communications. To decrypt
C=<U, V> using the escrow key s, compute: V.sym.H(
(U,sP.sub.pub))=M. [0209] As shown in FIG. 9, the escrow obtains
the system parameters and element P in block 900, then in block 910
obtains the recipient's public key xP, and obtains the encrypted
message V and encapsulated key rP from the sender. The escrow then
computes the message key in block 920, and uses it to decrypt the
message in block 930. The escrow can compute the message key from
the knowledge of s, rP, and xP. A standard argument shows that
assuming that BDH is hard for groups generated by the system of
this embodiment has semantic security in the random oracle model
(recall that since DDH is easy we cannot prove semantic security
based on DDH). Yet, the escrow agent can decrypt any ciphertext
encrypted using any user's public key. The decryption capability of
the escrow agent can be distributed using the PKG distribution
techniques described earlier.
[0210] Another embodiment uses a similar hardness assumption, with
an ElGamal encryption system with non-global escrow. In this
embodiment, each user constructs a public key with two
corresponding private keys, and gives one of the private keys to
the trusted third party. The trusted third party maintains a
database of all private keys given to it by the various users.
Although both private keys can be used to decrypt, only the user's
private key can be used simultaneously as the signing key for a
discrete logarithm based signature scheme.
[0211] Various other cryptographic systems can be devised based on
the principles illustrated in the above embodiments. For example,
three entities A, B, and C can communicate securely as a group by
privately selecting random integers a, b, c and publishing public
keys aP, bP, cP. One of them, such as A, could encrypt a message
using the message key (bP, cP).sup.r and transmit it with rP. Then
B could decrypt the message by calculating (cP, rP).sup.b and C
could decrypt it by calculating (bP, rP).sup.c. Similarly, B could
send a message to A and C, or C could send a message to A and
B.
[0212] In another possible embodiment, two of the three entities,
say A and B, could publish a joint public key abP. Then C could
encrypt a message using the message key (abP, cP).sup.r and
transmit it with rP. Then neither A nor B alone could decrypt the
message, but both A and B together could compute (cP, rP).sup.ab
and jointly decrypt the message. This technique generalizes to any
number of entities. For example, C could join A and B by using abP
to compute and publish the three-way joint public key abcP. Then
anyone could encrypt a message using the message key (abcP,
xP).sup.r and transmit it with rP. Then only A and B and C together
could compute (xP, rP).sup.abc and jointly decrypt the message.
Threshold Decryption.
[0213] Embodiments of the invention enable n entities to have
shares of a private key d.sub.ID corresponding to a given public
key ID, so that messages encrypted using ID can only be decrypted
if t of the n entities collaborate. The private key d.sub.ID is
never reconstructed in a single location. Embodiments of our IBE
system may support this as follows.
[0214] Recall that in other embodiments the private key
d.sub.ID=sQ.sub.ID where s.epsilon.*.sub.q is the master-key.
Instead, let s.sub.1, . . . s.sub.n.epsilon.*.sub.q be a t-out-of-n
Shamir secret sharing of the master-key s. Each of the n users is
given d.sub.i=s.sub.iQ.sub.ID. To decrypt a ciphertext <U, V>
encrypted using the key ID each user locally computes g.sub.i= (U,
d.sub.i) and sends g.sub.i.epsilon..sub.2 to the user managing the
decryption process. That user then combines the decryption shares
by computing g.sub.ID= i .times. g i .lamda. i ##EQU9## where
.lamda..sub.i are the appropriate Lagrange interpolation
coefficients used in Shamir secret sharing. The message is then
obtained by computing H.sub.2(g.sub.ID).sym.V=M.
[0215] Those skilled in the art of cryptography will be able to
devise many other schemes that employ the basic principles of the
present invention.
Applications of Identity-Based Encryption
[0216] One application for embodiments of identity-based encryption
is to help the deployment of a public key infrastructure. In this
section, we show several embodiments of this and other
applications.
Revocation of Public Keys
[0217] In this embodiment, the sender may encrypt using a public
key derived from a piece of information containing a time element,
such as a year, date or other time, to help provide key expiration
or other forms of temporal key management. For example, in one
embodiment, key expiration can be done by having Alice encrypt
e-mail sent to Bob using the public key: "bob@company.com
.parallel. current-year". In doing so Bob can use his private key
during the current year only. Once a year Bob needs to obtain a new
private key from the PKG. Hence, we get the effect of annual
private key expiration. Note that unlike the existing public key
infrastructure, Alice does not need to obtain a new certificate
from Bob every time Bob refreshes his private key.
[0218] One may make this approach more granular in other
embodiments by encrypting e-mail for Bob using "bob@company.com
.parallel. current-date", or using another time stamp. This forces
Bob to obtain a new private key every day. This embodiment may be
used in a corporate context where the PKG is maintained by the
corporation. With this approach key revocation is very simple: when
Bob leaves the company and his key needs to be revoked, the
corporate PKG is instructed to stop issuing private keys for Bob's
e-mail address. As a result, Bob can no longer read his email. The
interesting property is that Alice does not need to communicate
with any third party certificate directory to obtain Bob's daily
public key. Hence, embodiments of identity based encryption can
provide a very efficient mechanism for implementing ephemeral
public keys. Also note that this embodiment can be used to enable
Alice to send messages into the future: Bob will only be able to
decrypt the e-mail on the date specified by Alice.
Managing User Credentials
[0219] An embodiment of the invention enables the management of
user credentials using an IBE system. The message is encrypted with
a string containing a credential identifier. For example, suppose
Alice encrypts mail to Bob using the public key: "bob@company.com
.parallel. current-year .parallel. clearance=secret". Then Bob will
only be able to read the email if on the specified date he has
secret clearance. Consequently, it is very easy to grant and revoke
user credentials using the PKG.
[0220] FIG. 10 is a block diagram illustrating a system for
managing credentials in an identity based encryption system
according to an embodiment of the invention. The system includes
sender system 1001, receiver system 1002 and PKG 1003. Each such
system may be implemented as a computer system such as a client or
server connected to a computer network. Accordingly, sender 1001,
receiver 1002 and PKG 1003 may each contain processors, such as
processor 1014, processor 1013 and processor 1012. Additionally,
these systems may include computer-readable storage media, such as
computer memory, and may additionally include interfaces to a
computer network, including technology allowing for communication
with a wired, wireless or other network. Sender system 1001 may
include a software plug-in 1017. Such a plug-in may comprise a
software module which performs cryptographic functions. The plug-in
includes, according to an embodiment of the invention, items such
as cryptographic logic 1004. Plug-in 1017 may be distributed to
various computers such as sender system 1001 and receiver system
1002 through a network in order to roll out functionality
associated with identity-based encryption and other communication
functionality. Parameters 1015 from a system such as PKG 1003 are
also distributed over a computer network or other communications
medium to senders and receivers, such as sender system 1001 and
receiver system 1002, who may then use them in conjunction with
plug-in 1017 when encrypting or decrypting messages. In one
embodiment, plug-in 1017 is distributed together with parameters
1014. In an alternate embodiment, parameters 1015 may be
distributed separately.
[0221] Sender system 1001 encrypts a message M using encryption
logic 1004 in plug-in 1017. Encryption logic 1004 encrypts the
message using encryption key 1011, which is based on selected
credential 1005 and an identification 1016 of the intended receiver
of the message. In some embodiments, the key may be based on other
information as well. The sender system 1001 sends the receiver
system 1002 information 1006, e.g., in the form of a data packet
transmitted over a network or other communication medium. The
information 1006 sent to receiver system 1002 contains the
encrypted message and may also contain information 1007 regarding
the credential 1005 used as part of the basis for the encryption
key.
[0222] Either before or after receiving information 1006, receiver
system 1002 sends a request 1009 to PKG 1003. In one embodiment,
the request 1009 may include the receiver's identity 1016 and may
also include information related to the selected credential 1005.
In response, PKG 1003 verifies the credential of receiver 1002
using credential check logic 1008. Such logic may be implemented in
software, hardware or a combination thereof. If the credential is
verified as belonging to the receiver, then PKG 1003 provides a
response 1010 to receiver 1002, which includes a private decryption
key 1018 corresponding to the encryption key 1011. Using the
private decryption key, the receiver then may decrypt the encrypted
message contained in information 1006 to recover the original
message M. Thus, by including a credential as part of an encryption
key, embodiments such as this one allow a sender to encrypt a
message intended for a receiver, where the decryption of the
message by the receiver is contingent upon the validity of the
receiver's credential.
Delegation of Decryption Keys
[0223] Another application for embodiments of IBE systems is
delegation of decryption capabilities. We give two exemplary
embodiments, described with reference to a user Bob who plays the
role of the PKG. Bob runs the setup algorithm to generate his own
IBE system parameters params and his own master-key. Here we view
params as Bob's public key. Bob obtains a certificate from a CA for
his public key params. When Alice wishes to send mail to Bob she
first obtains Bob's public key params from Bob's public key
certificate. Note that Bob is the only one who knows his master-key
and hence there is no key-escrow with this setup. [0224] 1.
Delegation to a laptop. Suppose Alice encrypts mail to Bob using
the current date as the IBE encryption key (she uses Bob's params
as the IBE system parameters). Since Bob has the master-key he can
extract the private key corresponding to this IBE encryption key
and then decrypt the message. Now, suppose Bob goes on a trip for
seven days. Normally, Bob would put his private key on his laptop.
If the laptop is stolen the private key is compromised. When using
the IBE system Bob could simply install on his laptop the seven
private keys corresponding to the seven days of the trip. If the
laptop is stolen, only the private keys for those seven days are
compromised. The master-key is unharmed. FIG. 11 is a block diagram
illustrating a system with key delegation according to an
embodiment of the invention. The system includes user system 1101
and target system 1102. The target system may comprise a computer
such as a laptop computer. User system 1101 includes a master key
1103, which is used to generate decryption keys 1104. The
decryption keys 1104 are downloaded to the target system 1102.
Using the techniques of key revocation described above, these
decryption keys may be valid only for a limited time, thus
providing additional security in the event that target system 1101
is compromised. User system 1101 and target system 1102 may include
elements of computer systems such as memory 1106 and 1107 as well
as processor 1105 and 1108. User system 1101 includes key generator
logic 1109, which uses master key 1103 and system parameters 1110
to generate private decryption keys 1104 based on information
derived from a user ID 1113 and one or more dates 1114 or other
time stamps. Target system 1102 includes decryption logic 1111,
which uses the private decryption keys 1104 obtained from user
system 1101 and system parameters 1110 to decrypt an incoming
encrypted message 1112. If message 1112 is encrypted using public
keys based on ID 1113 and one of the dates 1114, then private
decryption keys may be used to decrypt it. Thus the decryption
capabilities of target system 1102 may be limited to messages
associated with selected dates 1114. In an alternate embodiment,
the target system may be a data storage medium or portable data
storage device which can be connected as desired to other computer
systems, thereby enabling use of the decryption keys on those
systems. [0225] 2. Delegation of duties. Suppose Alice encrypts
mail to Bob using the subject line as the IBE encryption key. Bob
can decrypt mail using his master-key. Now, suppose Bob has several
assistants each responsible for a different task (e.g. one is
`purchasing`, another is `human-resources`, etc.). In this
embodiment, Bob may give one private key to each of his assistants
corresponding to the assistant's responsibility. Each assistant can
then decrypt messages whose subject line falls within its
responsibilities, but it cannot decrypt messages intended for other
assistants. Note that Alice only obtains a single public key from
Bob (params), and she uses that public key to send mail with any
subject line of her choice. The mail can only be read by the
assistant responsible for that subject. [0226] More generally,
embodiments of IBE can simplify various systems that manage a large
number of public keys. Rather than storing a big database of public
keys the system can either derive these public keys from user
names, or simply use the integers 1, . . . , n as distinct public
keys. For example, in a corporation, each employee might have a
unique employee number, and that number may serve also as the
employee's public key. Return Receipt
[0227] FIG. 12 is a block diagram illustrating an encryption system
with return receipt according to an embodiment of the invention.
According to one embodiment of the invention, a sender can receive
an confirmation that the recipient has received an encrypted
message. More generally, upon receipt of a request for a decryption
key from a receiver, the PKG takes an action separate from
providing a decryption key to the receiver. Such an action
comprises providing an acknowledgement to the sender that indicates
that the message was received, according to one embodiment.
[0228] An embodiment of a system having return receipt capability
is illustrated in FIG. 12. The system includes sender system 1201,
recipient system 1202 and PKG system 1203. The sender system 1201,
receiver system 1202 and PKG system 1203 may be implemented as
computer systems coupled to a computer network. For example, PKG
1203, sender system 1201 and receiver system 1202 may include
processor 1212, processor 1213 and processor and 1214,
respectively. These computer systems may include elements such as
computer readable storage media, computer memory and other storage
devices. Additionally, these systems may include interfaces to a
computer network, including technology allowing for communication
from a wired, wireless or other network. Further, according an
embodiment of the invention, communication between the respective
elements may take place using data packets sent over a computer
network, or using any of various other forms of electronic and data
transmission and communication.
[0229] The sender 1201 encrypts a message M and sends the resulting
ciphertext to receiver 1202 in a data package 1204 that also may
include return receipt request information 1209. The return receipt
request information may contain, for example, a return address and
a message identifier corresponding to the particular message 1204.
The message M is encrypted by the sender using encryption logic
1211 and an encryption key 1215. Encryption key 1215 may be based
on a receiver ID (such as an e-mail address) 1216 and the return
receipt request information 1209. Because the receiver ID and
return receipt request information 1209 are used by the sender to
determine the encryption key 1215, the receiver 1202 needs a
corresponding decryption key that can be used to decrypt the
message. Accordingly, recipient system 1202, in response to
receiving message 1204, sends PKG 1203 a request 1206, which
includes the return receipt request information 1209 and the
receiver's ID, 1216. In response, PKG 1203 sends to receiver 1202
the private decryption key 1205, which receiver then uses with
decryption logic 1217 to decrypt the ciphertext of message 1204 and
recover the original message M. In addition to sending receiver
1202 the decryption key 1205, PKG 1203 also sends a return receipt
1207 to sender 1201. PKG 1203 may alternatively store the receipt
on storage media as part of a log rather than send a return
receipt. Return receipt 1207 may include information such as the
message identifier. Thus, sender 1201 receives proof that recipient
1202 has received the message 1204. The system may be initialized
by placing plug-in software in various systems, such as sender
system 1201 and receiver system 1202. Such plug-in software may
include system parameters, some of which may be derived from a
system master key. Such parameters, stored in local devices such as
sender 1201 and receiver 1202 are then used to generate encryption
keys, perform encryption, perform decryption, and other functions,
as appropriate.
Description of the Weil Pairing
[0230] In this section we describe the Weil pairing on elliptic
curves and then show how to efficiently compute it using an
algorithm. To be concrete we present an example using supersingular
elliptic curves defined over a prime field .sub.p with p>3 (the
curve y.sup.2=x.sup.3+1 over .sub.p with p=2 mod 3 is an example of
such a curve). The following discussion easily generalizes to
computing the Weil pairing over other elliptic curves.
Elliptic Curves and the Weil Pairing
[0231] We state a few elementary facts about supersingular elliptic
curves defined over a prime field .sub.p with p>3: [0232] Fact
1: A supersingular curve E/.sub.p (with p>3) contains p+1 points
in .sub.p. We let O denote the point at infinity. The group of
points over .sub.p forms a cyclic group of order p+1. For
simplicity, let P be a generator of this group and set n=p+1.
[0233] Fact 2: The group of points E(.sub.p.sub.2) contains a point
Q of order n which is linearly independent of the points in
E(.sub.p). Hence, E(.sub.p.sub.2) contains a subgroup which is
isomorphic to the group .sub.n.sup.2. The group is generated by
P.epsilon.E(.sub.p) and Q.epsilon.E(.sub.p.sub.2). We denote this
group by E[p+1]=E[n]. We will be working with the Weil pairing e
which maps pairs of points in E[n] into *.sub.p.sub.2, i.e. e:
E[n].times.E[n].fwdarw.*.sub.p.sub.2. To describe the pairing, we
review the following concepts: [0234] Divisors A divisor is a
formal sum of points on the curve E(.sub.p.sub.2). We write
divisors as = P .times. a p .function. ( P ) ##EQU10## [0235] where
a.sub.P.epsilon. and P.epsilon.E(.sub.p.sub.2). For example,
=3(P.sub.1)-2(P.sub.2)-(P.sub.3) is a divisor. We will only
consider divisors f .function. ( x , y ) .di-elect cons. p 2
.function. ( x , y ) . ##EQU11## [0236] where P .times. a p = 0.
##EQU12## [0237] Functions Roughly speaking, a function f on the
curve E(.sub.p.sub.2) can be viewed as a rational function f(x,
y).epsilon..sub.p.sub.2(x, y). For any point P=(x,
y).epsilon.E(.sub.p.sub.2) we define f(P)=f(x, y). [0238] Divisors
of functions Let f be a function on the curve E(.sub.p.sub.2). We
define its divisor, denoted by (f), as ( f ) = P .times. ord P
.function. ( f ) P . ##EQU13## [0239] P. Here ord.sub.P(f) is the
order of the zero that f has at the point P. For example, let
ax+by+c=0 be the line passing through the points P.sub.1,
P.sub.2.epsilon.E(.sub.p.sub.2) with P.sub.1 .noteq..+-.P.sub.2.
This line intersects the curve at third point
P.sub.3.epsilon.E(.sub.p.sub.2). Then the function f(x, y)=ax+by+c
has three zeroes P.sub.1, P.sub.2, P.sub.3 and a pole of order 3 at
infinity. The divisor of f is
(f)=(P.sub.1)+(P.sub.2)+(P.sub.3)-3(O). [0240] Principal divisors
Let be a divisor. If there exists a function f such that (f)= then
we say that is a principal divisor. We know that a divisor = P
.times. a p .function. ( P ) ##EQU14## [0241] is principal if and
only if P .times. a p = 0 ##EQU15## [0242] and P .times. a P
.times. P = O . ##EQU16## [0243] Note that the second summation is
using the group action on the curve. Furthermore, given a principal
divisor there exists a unique function f (up to constant multiples)
such that (A)=(f). [0244] Equivalence of divisors We say that two
divisors , are equivalent if their difference - is a principal
divisor. We know that any divisor = P .times. a p .function. ( P )
##EQU17## [0245] (with ( with .times. .times. P .times. a P = 0 )
##EQU18## [0246] is equivalent to a divisor of the form '=(Q)-(O)
for some Q.epsilon.E. Observe that Q = P .times. a P .times. P .
##EQU19## [0247] Notation Given a function f and a divisor = P
.times. a P .function. ( P ) ##EQU20## [0248] we define f() as
f()=.PI..sub.Pf(P).sup.aP. Note that since P .times. a P = 0
##EQU21## [0249] we have that f() remains unchanged if instead of f
we use cf for any c.epsilon..sub.p.sub.2. We are now ready to
describe the Weil pairing of two points P, Q.epsilon.E[n]. Let
.sub.P be some divisor equivalent to the divisor (P)-(O). We know
that n.sub.P is a principal divisor (it is equivalent to n(P)-n(O)
which is clearly a principal divisor). Hence, there exists a
function f.sub.P such that (f.sub.P)=n.sub.P. Define .sub.Q and
f.sub.Q analogously. The Weil pairing of P and Q is given by: e
.function. ( P , Q ) = f P .function. ( Q ) f Q .function. ( P )
##EQU22## This ratio provides the value of the Weil pairing of P
and Q whenever it is well defined (i.e., whenever no division by
zero has occurred). If this ratio is undefined we use different
divisors .sub.P, .sub.Q to define e(P, Q). When P,
Q.epsilon.E(.sub.p.sub.2) we have that e(P,
Q).epsilon..sub.p.sub.2.
[0250] We briefly show that the Weil pairing is well defined. That
is, the value of e(P, Q) is independent of the choice of the
divisor .sub.P as long as .sub.P is equivalent to (P)-(O) and
.sub.P leads to a well defined value. The same holds for .sub.Q.
Let {circumflex over ()}.sub.P be a divisor equivalent to .sub.P
and let {circumflex over (f)}.sub.P be a function so that
({circumflex over (f)}.sub.P)=n{circumflex over ()}.sub.P. Then
{circumflex over ()}.sub.P=.sub.P+(g) for some function g and
{circumflex over (f)}.sub.P=f.sub.Pg.sup.n. We have that: e
.function. ( P , Q ) = f ^ P .function. ( Q ) f Q .function. ( ^ P
) = f P .function. ( Q ) .times. g .function. ( Q ) n f Q
.function. ( P ) .times. f Q .function. ( ( g ) ) = f P .function.
( Q ) f Q .function. ( P ) g .function. ( n .times. .times. Q ) f Q
.function. ( ( g ) ) = f P .function. ( Q ) f Q .function. ( P ) g
.function. ( ( f Q ) ) f Q .function. ( ( g ) ) = f P .times. ( Q )
f Q .times. ( P ) ##EQU23## The last equality follows from the
following fact known as Weil reciprocity: for any two functions f,
g we have that f((g))=g((f)). Hence, the Weil pairing is well
defined. Fact 10 The Weil pairing has the following properties:
[0251] For all P.epsilon.E[n] we have: e(P, P)=1. [0252] Bilinear:
e(P.sub.1+P.sub.2, Q)=e(P.sub.1, Q)e(P.sub.2, Q) and e(P,
Q.sub.1+Q.sub.2)=e(P, Q.sub.1)e(P, Q.sub.2) [0253] When P,
Q.epsilon.E[n] are collinear then e(P, Q)=1. Similarly, e(P,
Q)=e(Q, p).sup.-1. [0254] n'th root: for all P, Q.epsilon.E[n] we
have e(P, Q).sup.n=1. [0255] Non-degenerate: if P satisfies e(P,
Q)=1 for all Q.epsilon.E[n] then P=O.
[0256] As discussed earlier, our detailed example of an embodiment
of an IBE scheme uses the modified Weil pairing (P, Q)=e(P,
.phi.(Q)), where .phi. is an automorphism on the group of points of
E. Tate pairing. The Tate pairing is another bilinear pairing that
has the required properties for embodiments of our system. In
various embodiments, we slightly modify the original definition of
the Tate pairing to fit our purpose. Define the Tate pairing of two
points P, Q.epsilon.E[n] as T .function. ( P , Q ) = f P .function.
( Q ) p 2 * / n ##EQU24## where f.sub.P and .sub.Q are defined as
earlier. This definition gives a computable bilinear pairing T:
E[n].times.E[n].fwdarw..sub.2. Computing the Weil Pairing
[0257] Given two points P, Q.epsilon.E[n] we show how to compute
e(P, Q).epsilon.*.sub.p.sub.2 using O(log p) arithmetic operations
in .sub.p. We assume P # Q. We proceed as follows: pick two random
points R.sub.1, R.sub.2.epsilon.E[n]. Consider the divisors
.sub.P=(P+R.sub.1)-(R.sub.1) and .sub.Q=(Q+R.sub.2)-(R.sub.2).
These divisors are equivalent to (P)-(O) and (O)-(O) respectively.
Hence, we can use .sub.P and .sub.Q to compute the Weil pairing as:
e .function. ( P , Q ) = f P .times. ( Q ) f Q .times. ( P ) = f P
.function. ( Q + R 2 ) .times. f Q .function. ( R 1 ) f P
.function. ( R 2 ) .times. f Q .function. ( P + R 1 ) ##EQU25##
This expression is well defined with very high probability over the
choice of R.sub.1, R.sub.2 (the probability of failure is at most O
.function. ( log .times. .times. p p ) ) . ##EQU26## In the rare
event that a division by zero occurs during the computation of e(P,
Q) we simply pick new random points R.sub.1, R.sub.2 and repeat the
process.
[0258] To evaluate e(P, Q) it suffices to show how to evaluate the
function f.sub.P at .sub.Q. Evaluating f.sub.Q(.sub.P) is done
analogously. We evaluate f.sub.P(.sub.Q) using repeated doubling.
For a positive integer b define the divisor
.sub.b=b(P+R.sub.1)-b(R.sub.1)-(bP)+(O) It is a principal divisor
and therefore there exists a function f.sub.b such that
(f.sub.b)=.sub.b. Observe that (f.sub.P)=(f.sub.n) and hence,
f.sub.P(.sub.Q)=f.sub.n(.sub.Q). It suffices to show how to
evaluate f.sub.n(.sub.Q). Lemma 11 There is an algorithm that given
f.sub.b(.sub.Q), f.sub.c(.sub.Q) and bP, cP, (b+c)P for some b,
c>0 outputs f.sub.b+c(.sub.Q). The algorithm only uses a (small)
constant number of arithmetic operations in .sub.p.sub.2. Proof We
first define two auxiliary linear functions g.sub.1, g.sub.2:
[0259] 1. Let a.sub.1x+b.sub.1y+c.sub.1=0 be the line passing
through the points bP and cP (if b=c then let
a.sub.1x+b.sub.1y+c.sub.1=0 be the line tangent to E at bP). Define
g.sub.1(x, y)=a.sub.1x+b.sub.1y+c.sub.1. [0260] 2. Let x+c.sub.2=0
be the vertical line passing through the point (b+c)P. Define
g.sub.2(x, y)=x+c.sub.2 The divisors of these functions are:
(g.sub.1)=(bP)+(cP)+(-(b+c)P)-3(O)
(g.sub.2)=((b+c)P)+(-(b+c)P)-2(O) By definition we have that:
.sub.b=b(P+R.sub.1)-b(R.sub.1)-(bP)+(O)
.sub.c=c(P+R.sub.1)-c(R.sub.1)-(cP)+(O)
.sub.b+c=(b+c)(P+R.sub.1)-(b+c)(R.sub.1)-((b+c)P)+(O) It now
follows that: .sub.b+c=.sub.b+.sub.c+(g.sub.1)-(g.sub.2). Hence: f
b + c .function. ( Q ) = f b .function. ( Q ) f c .function. ( Q )
g 1 .function. ( Q ) g 2 .function. ( Q ) ( 1 ) ##EQU27## This
shows that to evaluate f.sub.b+c(.sub.Q) it suffices to evaluate
g.sub.i(.sub.Q) for all i=1, 2 and plug the results into equation
1. Hence, given f.sub.b(.sub.Q), f.sub.c(.sub.Q) and bP, cP, (b+c)P
one can compute f.sub.b+c(.sub.Q) using a constant number of
arithmetic operations.
[0261] Denote the output of Algorithm of Lemma 11 by
(f.sub.b(.sub.Q), f.sub.c(.sub.Q), bP, cP,
(b+c)P)=f.sub.b+c(.sub.Q). Then one can compute
f.sub.P(.sub.Q)=f.sub.n(.sub.Q) using the following standard
repeated doubling procedure. Let n=b.sub.mb.sub.m-1 . . .
b.sub.1b.sub.0 be the binary representation of n, i.e. n = i = 0 m
.times. b i .times. 2 i . ##EQU28## Init: Set Z=O,
V=f.sub.0(.sub.Q)=1, and k=0. Iterate: For i=m, m-1, . . . , 1, 0
do: [0262] 1: If b.sub.i=1 then do: Set V=(V, f.sub.1(.sub.Q), Z,
P, Z+P), set Z=Z+P, and set k=k+1. [0263] 2: If i>0 set V=(V, V,
Z, Z, 2Z), set Z=2Z, and set k=2k. [0264] 3: Observe that at the
end of each iteration we have z=kP and V=f.sub.k(.sub.Q). Output:
After the last iteration we have k=n and therefore
V=f.sub.n(.sub.Q) as required. To evaluate the Weil pairing e(P, Q)
we run the above algorithm once to compute f.sub.P(.sub.Q) and once
to compute f.sub.Q(.sub.P). Note that the repeated squaring
algorithm needs to evaluate f.sub.1(3 .sub.Q). This is easily done
since the function f.sub.1(x, y) (whose divisor is
(f.sub.1)=(P+R.sub.1)-(R.sub.1)-(P)+(O)) can be written out
explicitly as follows: [0265] 1. Let a.sub.1x+b.sub.1y+c.sub.1=0 be
the line passing through the points P and R.sub.1. Define the
function: g.sub.1(x, y)=a.sub.1x+b.sub.1y+c.sub.1. [0266] 2. Let
x+c.sub.2=0 be the vertical line passing through the point
P+R.sub.1. Define the function: g.sub.2(x, y)=x+c.sub.2. [0267] 3.
The function f.sub.1(x, y) is simply f.sub.1(x, y)=g.sub.2(x,
y)/g.sub.1(x, y) which is easy to evaluate in .sub.p.sub.2.
* * * * *