U.S. patent application number 11/788795 was filed with the patent office on 2014-12-18 for system and method for analyzing unauthorized intrusion into a computer network.
The applicant listed for this patent is Alen Capalik. Invention is credited to Alen Capalik.
Application Number | 20140373144 11/788795 |
Document ID | / |
Family ID | 38950746 |
Filed Date | 2014-12-18 |
United States Patent
Application |
20140373144 |
Kind Code |
A9 |
Capalik; Alen |
December 18, 2014 |
System and method for analyzing unauthorized intrusion into a
computer network
Abstract
The method analyzes unauthorized intrusion into a computer
network. Access is allowed through one or more open ports to one or
more virtualized decoy operating systems running on a hypervisor
operating system hosted on a decoy network device. This may be done
by opening a port on one of the virtualized decoy operating
systems. A network attack on the virtualized operating system is
then intercepted by an introspection module running on the
hypervisor operating system. The attack-identifying information is
communicated through a private network interface channel and stored
on a database server as forensic data. A signature-generation
engine uses this forensic data to generate a signature of the
attack. An intrusion prevention system then uses the attack
signature to identify and prevent subsequent attacks. A web-based
visualization interface facilitates configuration of the system and
analysis of (and response to) forensic data generated by the
introspection module and the signature generation engine, as well
as that stored in the processing module's relational databases.
Inventors: |
Capalik; Alen; (Los Angeles,
CA) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Capalik; Alen |
Los Angeles |
CA |
US |
|
|
Prior
Publication: |
|
Document Identifier |
Publication Date |
|
US 20080016570 A1 |
January 17, 2008 |
|
|
Family ID: |
38950746 |
Appl. No.: |
11/788795 |
Filed: |
April 20, 2007 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11488743 |
Jul 17, 2006 |
8429746 |
|
|
11788795 |
|
|
|
|
60802543 |
May 22, 2006 |
|
|
|
Current U.S.
Class: |
726/23 |
Current CPC
Class: |
H04L 63/1491 20130101;
G06F 21/552 20130101; G06F 21/566 20130101; G06F 9/45533 20130101;
H04L 63/145 20130101; H04L 63/1416 20130101; H04L 63/1408 20130101;
G06F 2221/2123 20130101 |
Class at
Publication: |
726/23 |
International
Class: |
G06F 12/14 20060101
G06F012/14 |
Claims
1. A method for analyzing unauthorized intrusion into a computer
network, the method comprising: allowing access to an apparently
vulnerable virtualized decoy operating system running on a
hypervisor operating system hosted on a decoy network device; using
an introspection module comprising a virtual-machine-based rootkit
module and its associated userland processes running on the
hypervisor operating system to intercept a network attack on the
virtualized operating system, wherein the network attack includes
attack-identifying information; and generating forensic data on the
network attack from the attack-identifying information.
2. The method of claim 1, further comprising: generating an attack
signature from the forensic data; and providing the attack
signature to an intrusion prevention system configured to control
access to a protected network using the attack signature to
identify subsequent attacks.
3. The method of claim 2, further comprising controlling access to
the protected network using the attack signature.
4. The method of claim 2, wherein the attack signature is
automatically generated by the system without human
intervention.
5. The method of claim 1, further comprising, before the allowing,
opening a port on the virtualized decoy operating system through
which the network attack is made.
6. The method of claim 1, further comprising: allowing access to an
additional virtualized decoy operating system running on the
hypervisor operating system; using the virtual-machine-based
rootkit module to intercept an additional network attack on the
additional virtualized operating system, wherein the additional
network attack includes additional attack-identifying information;
and generating additional forensic data on the additional network
attack from the additional attack-identifying information.
7. The method of claim 6, further comprising: generating an
additional attack signature from the additional forensic data; and
providing the additional attack signature to an intrusion
prevention system configured to control access to a protected
network using the attack-identifying information.
8. The method of claim 1, wherein the virtualized decoy operating
system and the additional virtualized decoy operating system are
different types of operating systems.
9. The method of claim 1, wherein said forensic data is generated
based on an attack payload including keystrokes, ASCII, or binary
files.
10. The method of claim 1, wherein the attack forensics are
generated if an attacker is able to successfully gain access to the
virtualized decoy operating system.
11. The method of claim 1, further comprising providing a report of
said network attack to a network administrator.
12. The method of claim 1, further comprising storing the attack
forensics in a database.
13. A system for analyzing unauthorized intrusion into a computer
network, the system comprising: a virtualized operating system
module comprising: a hypervisor operating system comprising: at
least one virtualized decoy operating system; a
virtual-machine-based rootkit configured to intercept a network
attack on the virtualized operating system, wherein the network
attack includes transmission of attack-identifying information; and
a processing module electrically coupled to the introspection
module via a network interface communication channel, wherein the
processing module comprises: a database configured to store
forensic data on the network attack.
14. The system of claim 13, wherein the processing module further
comprises an attack signature-generation engine configured to
generate an attack signature from the forensic data on the network
attack, wherein attack signatures may be generated on a timescale
ranging from near-immediate to several minutes after initiation of
an attack.
15. The system of claim 14, wherein the processing module further
comprises a web-based visualization interface that facilitates
configuration of the system and forensic analysis of captured
attack information by administrators.
16. The system of claim 15, further comprising an intrusion
prevention system electrically coupled to the signature-generation
engine.
17. The system of claim 13, wherein the intrusion detection system
is configured to prevent unauthorized intrusion into a protected
computer network.
18. The system of claim 13, wherein the network interface
communication channel is a private channel.
19. The system of claim 13, wherein the virtualized operating
system module includes multiple virtualized decoy operating systems
on the hypervisor operating system.
20. The system of claim 13, wherein the attack forensics are based
on an attack payload including keystrokes or ASCII or binary
files.
21. The system of claim 13, wherein the virtualized operating
system module and the processing module are contained in memory on
the same or separate computing devices that each includes a
processor.
22. A computing device configured for analyzing unauthorized
intrusion into a computer network, the device comprising: a
processor; and memory coupled to the processor, wherein the memory
comprises procedures for: allowing access to a virtualized decoy
operating system running on a hypervisor operating system hosted on
a decoy network device; using an introspection module running on
the hypervisor operating system to intercept a network attack on
the virtualized operating system, wherein the network attack
includes attack-identifying information; and generating forensic
data on the network attack from the attack-identifying
information.
23. The computing device of claim 22, further comprising a
web-based visualization module comprising procedures for: analyzing
forensic data generated by the introspection module and the
signature generation engine, as well as that stored in the
processing module's relational databases; and configuring the
system, wherein an administrator can tune the system's behavior,
including its pattern matching facilities, as well as automate the
system's response to attack-identifying information captured by the
introspection module and automate its response to forensic data
generated by the signature-generation engine and any information
stored on the processing module's relational databases.
Description
[0001] This is a Continuation-In-Part, and claims priority to and
the benefit of: U.S. Ser. No. 11/488,743, entitled "Decoy Network
Technology With Automatic Signature Generation for Intrusion
Detection and Intrusion Prevention Systems" filed on Jul. 17, 2006,
the entire disclosure of which is incorporated herein by
reference.
FIELD
[0002] The invention relates to the field of methods and systems
for protecting computer networks and is more particularly, but not
by way of limitation, directed to decoy network technology with
automatic signature generation for intrusion detection and
intrusion prevention systems.
BACKGROUND
[0003] Computer networks typically interface with the Internet or
other public computer systems and are thus vulnerable to attacks,
unwanted intrusions and unauthorized access. One threat to networks
is the so-called zero-day attack that exploits security
vulnerabilities unknown to the system operators.
[0004] Conventional network security systems include a firewall
that generally prevents unauthorized access to the network or its
computers. Conventional systems also include intrusion detection
systems (IDS) and intrusion prevention systems (IPS) that typically
contain a library of signatures of malware payloads, which enable
them to detect those defined exploits attempting to access
production systems. When a connection is attempted to a network
port, the IDS or IPS examines the low-level IP data packets and
compares them to its library of signatures for a match. When a
match is identified the IDS or IPS provides notification of the
match.
[0005] The problem lies in the static nature of the conventional
IDS and IPS signatures coupled with the ability of determined
attackers to launch new undefined or zero-day automated attacks to
gain access to the network. While an intrusion prevention system
(IPS) equipped with behavioral signatures providing the ability to
capture behavioral patterns offers a higher level of protection,
these have similar drawbacks in that behavioral signatures are
still static in nature and limited in their ability to stop
zero-day attacks.
[0006] Still another type of network security systems utilizes a
honeynet arrangement to attract and then trap a suspected attacker.
A honeynet is made up of two or more honeypots on a network. Such
measures typically are made up of a computer, data or network site
that appears to be part of the network and appears to be one or
more valuable targets, but which is actually an isolated component
located away from production networks. These are typically passive
measures effective against spammers and other low-level attacks.
Such systems typically run emulated operating systems and services
and are generally not useful against sophisticated attackers who
can detect and effectively avoid the honeynet, never unloading
their zero-day attack or payload for the honeynet to capture and
analyze. Also, if the conventional honeynet configuration is not
sufficiently separated from the network system, an attacker can use
the honeynet to gain access to the network. Examples of emulated or
software based honeypots include "honeyd" which is a GPL licensed
daemon that is utilized to simulate network structures. Another
example of emulated software based honeypots include "mwcollect"
and "nepenthes" which are also released under the GPL license and
which are utilized to collect malware. The "mwcollect" and
"nepenthes" packages extract information on obtaining the malware
binaries from the exploit payload.
[0007] Because each of the problems and limitations discussed above
exist in the prior art devices and systems, there is a need for
methods and systems that adequately protect networks from new and
undefined attacks and that allow for real-time updates to a
network's library of attack signatures.
SUMMARY
[0008] One or more embodiments of the invention are directed to an
improved method and system for protecting computer networks. In one
embodiment, the invention comprises a modular decoy network
appliance, which runs fully functional operating systems on client
hardware modules. The modular arrangement comprises front-end fully
functional operating system modules and a separate processing
back-end module.
[0009] The front-end presents a standard fully functional operating
system, such as Windows.RTM. or a flavor of Linux.RTM., or Sun
Microsystems Solaris.RTM. that returns a standard operating system
fingerprint when it is scanned by tools that attackers typically
use to identify vulnerable systems. The attacker is thus lured into
accessing the identified operating system and running custom or
known exploits on that system.
[0010] The front-end module includes a sentinel kernel driver (or a
more generalized executable module) that is hidden from system
scanners as it is removed from kernel module listings or registry
in Windows. Thus, the kernel does not indicate the sentinel kernel
driver is running. The sentinel kernel driver monitors connections
to the operating system as well as activity on the operating system
and activity on services running on the operating system. When an
attacker connects to a port, the sentinel kernel driver captures
the data coming through the socket. Generally all relevant data
coming through the socket is captured. In most cases this means
whatever data is received as part of an incoming attack is captured
by the sentinel driver. Captured data is sent as a slew of common
UDP packets to the back end processing module over the fabric
network connection separate from the vulnerable front-end modules.
In this manner, there is no way for the intruder to know that his
or her communications with the operating system are being
analyzed.
[0011] The captured data, which contains the attack-identifying
information, is sent to the back-end or processing module though
the backplane fabric of the appliance using Layer 2 Ethernet
communication protocol. The processing module is separate and
independent from the client operating system modules and
communicates the processed information to security administrators
through a network port connected to the private and secure VLAN.
Unbeknownst to the intruder, the exploit is thus captured,
transferred and analyzed.
[0012] With the received data, the processing module generates a
report of the attack. The report consists of user-friendly
information that paints a picture of the attack for a system
administrator. This may include information on which sockets were
accessed, what happened at a particular socket, the key strokes
entered or bytes transferred to the port, what files were
transferred, registry changes, how the attack was run, what
happened on the primary network, on its servers or how the network
services were affected. The report may also include information on
the location of the attacker or the attacker's service provider.
Graphical representations of key information and interactive
mapping of the attack locales by region or country may be utilized
in one or more embodiments of the invention.
[0013] The processing module is used to generate an attack
signature by analyzing all the data passed through the socket. The
signature is generated by analyzing the attack payload including
the keystrokes or transferred bytes and any files uploaded to the
client operating system of an ASCII or binary nature. The files
uploaded are assumed to be of a malicious nature created to deliver
a malicious payload in the form of a compiled program or an
interpreted script. In the event that no malicious files are
uploaded to the operating system, the signature generation engine
analyzes all the keystrokes or bytes delivered through the socket
and creates a pattern signature which when applied to an IDS or IPS
system, enables the IDS or IPS systems to detect the attack if
repeated on production systems. Once generated, the attack
signatures can be viewed by a system administrator to determine the
appropriate course of action. The system administrator can instruct
the signature to be uploaded to the intrusion detection system
(IDS) or intrusion prevention system (IPS) for the protected
network where it is added to the IDS's or IPS's library of
signatures to protect production systems. In one or more
embodiments of the invention, the signature may be uploaded or
saved in a third party system that maintains all known exploits. In
this manner, other systems may be notified through secure channels
of an impending threat. For example, by transferring the signature
to a centralized server that communicates with multiple
installations, the intruder may be thwarted before attacking other
systems in other companies.
[0014] A production network's library of signatures can be updated
in real-time as the attacker modifies its illicit activity or a new
attack is launched. The embodiment can also maintain a database of
any and all attack signatures generated. Other and further
advantages will be disclosed and identified in the description and
claims and will be apparent to persons skilled in the art.
[0015] Another embodiment provides a system and method for
analyzing unauthorized intrusion into a computer network. Access is
allowed through one or more open ports to one or more virtualized
decoy operating systems running on a hypervisor operating system
hosted on a decoy network device. This may be done by opening a
port on one of the virtualized decoy operating systems. A network
attack on the virtualized operating system is then intercepted by a
virtual-machine-based rootkit module running on the hypervisor
operating system. The attack-identifying information is
communicated through a private network interface channel and stored
on a database server as forensic data. A signature generation
engine uses this forensic data to generate a signature of the
attack. An intrusion prevention system then uses the attack
signature to identify and prevent subsequent attacks
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] FIG. 1 illustrates a block diagram of an embodiment of the
system;
[0017] FIG. 2 illustrates a flow chart of an embodiment of the
processing that occurs on processing module 40;
[0018] FIG. 3 illustrates a human readable summary of an example
attack;
[0019] FIG. 4 illustrates an XML formatted attack signature
generated from the attack summarized in FIG. 3 for transmittal to
an IDS or IPS; and
[0020] FIG. 5 illustrates a block diagram of another embodiment of
the system.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
[0021] The following descriptions of embodiments of the invention
are exemplary, rather than limiting, and many variations and
modifications are within the scope and spirit of the invention.
Although numerous specific details are set forth in order to
provide a thorough understanding of the present invention, it will
be apparent to one of ordinary skill in the art, that embodiments
of the invention may be practiced without these specific details.
In other instances, well-known features have not been described in
detail in order to avoid unnecessarily obscuring the present
invention.
[0022] One or more embodiments of the invention are directed to an
improved method and system for protecting computer networks. One
embodiment is illustrated in FIG. 1, which illustrates attacker
activity 10 directed at protected computer network 20. As in a
typical attack, attack 10 is scanning for an open port on computer
network 20 in an attempt to make a connection and then access one
or more protected network devices 20a on network 20.
[0023] Attack 10 is monitored by decoy 100 that includes at least
one monitor/intercept module 30. Monitor/intercept module 30
comprises fully functioning decoy operating system 32 that monitors
each of the access ports for network 20. Any operating system may
be used as decoy operating system 32 including Windows.RTM., Sun
Microsystems Solaris.RTM. or any version of Linux.RTM. known to
persons skilled in the art. All known operating systems are within
the scope of the present invention. FIG. 1 shows one
monitoring/intercept module 30 in the foreground, however any
number of homogeneous or heterogeneous monitoring/intercept modules
may be utilized (shown as a stack behind monitor/intercept module
30). For example, in one embodiment of the invention a Windows.RTM.
monitoring/intercept module 30 and LINUX.RTM. monitoring/intercept
module 30 may be employed. There is no limit to the number of
monitoring/intercept modules that may be utilized in the system and
other embodiments may employ homogeneous decoy operating systems 32
that are of the same or of different versions. Monitoring/intercept
module 30 also includes sentinel kernel driver 34 which will be
described in further detail below. Protected network devices 20a
are accessed through IDS/IPS with Library of Signatures 62 in one
or more embodiments of the invention. The system also includes
processing module 40 for obtaining and analyzing exploits.
[0024] When attack 10 connects to an access port of network 20, the
fully functional decoy operating system 32 intercepts the
connection and returns a standard operating system fingerprint. For
example when connecting to an address that does not exist on
protected network 20, decoy 30 may be configured to respond to any
such incorrect address since the connection is assumed to be
malicious as there is no hardware on protected network 20 at that
address. The response may be configured to utilize any existing
hardware module having a given operating system and version within
monitoring/intercept module 30. For example, an FTP port access for
Windows.RTM. may return a particular character sequence that is
different than an FTP response for LINUX.RTM.. An FTP access to a
Windows.RTM. port for example may return a response ">ftp:
connect: Connection refused". This characters sequence may be
slightly different on LINUX.RTM. and hence allows the intruder to
determine what type of operating system is at a particular network
address. In addition, different versions of Windows.RTM. may
respond with slightly different character sequences which allows
the intruder to determine the specific version of the operating
system or to determine a possible range of versions for the
responding operating system. The instigator of attack 10 is thus
lured into accessing decoy 100, which includes monitor/intercept
module 30, and running custom or known exploits for the observed
operating system. When attacker activity proceeds to interact with
decoy 100, the attacker provides decoy 100 with the data used to
obtain control of decoy 100, which is recorded and analyzed without
knowledge of the attacker.
[0025] All scans by attack 10 receive real-world operating system
information, thus leading the instigator of the attack 10 to
believe that there is a potentially vulnerable system responding
and thus luring attack 10 into communicating with monitor/intercept
module 30. Since real hardware is utilized, the attacker is
attacking an actual physical system and thus has no idea that the
system is actually an instrumented honeypot that monitors the
attackers every move.
[0026] Monitor/intercept module 30 includes sentinel kernel driver
34. In one embodiment, sentinel kernel driver 34 is a combination
of custom root-kit code that on Windows.RTM. based operating
systems removes pointers from Microsoft.RTM. client/server runtime
server subsystem (CSRSS.exe). This coupled with removing sentinel
kernel driver 34 from the Windows.RTM. registry effectively hides
sentinel kernel driver 34 and all its drivers from attack 10. On
Unix.RTM. based operating systems, the kernel pointers are removed
making the kernel unable to link to a running process, effectively
hiding sentinel kernel driver 34 and all its libraries from attack
10. Sentinel kernel driver 34 monitors all data coming through the
socket and is derived from an open source code, such as libpcap,
known to persons skilled in the art.
[0027] When an attacker connects to a port, and begins interacting
with decoy operating system 32, sentinel 34 monitors and captures
information from the connection including port numbers, data
streams, keystrokes, file uploads and any other data transfers.
[0028] The captured information, or attack-identifying information,
is then sent for processing to processing module 40 as illustrated
in FIG. 1. Processing module 40 may optionally include a sentinel
server that receives information from the sentinel kernel driver
and deposits the information in a database for later analysis. In
one embodiment, the monitor/intercept module 30 is a front-end
module or series of modules and the captured data is sent to
processing module 40 though the backplane of the appliance or
appliances through a layer 2 Ethernet communications link not
available to the attacker such as an IP connection or any other
hardware dependent custom communication protocol known to persons
skilled in the art. Processing module 40 is part of a secure and
separate administrative network 42. In one or more embodiments the
signature may be sent from the back end processing module 40 to
IDS/IPS 62 through a second network connection which is used by the
processing module 40 to directly interact with IDS/IPS 62. The
sentinel kernel driver may utilize replay functionality to replay
the attacks on the operating system in reverse to clean up the
operating system to its pre-attack state. In this manner, the
attack can be thwarted and the operating system thus does not
become a tool of the hacker.
[0029] As shown in FIG. 2, processing starts at 200 and waits for
activity from sentinel kernel driver 34 at step 43. In step 44,
processing module 40 generates a report of the attack that includes
attack-identifying information (See FIG. 3). This report is for the
use and review by a system administrator who is responsible for
administering protected network 20. The attack may contain one or
more data transfers or keystrokes for example, which are analyzed
at step 46. By observing whether the attacker is successful in
interacting with the system, i.e., if the system is responding in a
manner that shows that the attacker has gained access, then
determination whether to generate an attack signature is made at
step 48 and the attack signature is generated at step 52 (See FIG.
4). If the attacker for example is unsuccessful at gaining access
or if there is no data transfer for example, then the attack
inquiry may be ended at step 50. Any generated attack signature is
sent to the IDS/IPS at step 56 and processing continues at step
43.
[0030] In one embodiment of the invention, the report is written,
and is displayed in an web-based visualization interface and can
include information about which sockets were accessed by attack 10,
what happened at a particular socket, the key strokes entered or
data transferred, what files were transferred, how the attack 10
was run, what happened on monitor/intercept module 30 and how decoy
operating system 32 and any related network services were affected.
The report may also include information on the location of the
instigator of attack 10 or the service provider used for attack 10.
Graphical representations of key information and interactive
mapping of attack locales by region or country may also be included
in the report.
[0031] In step 46, the attack-identifying information is analyzed
for known attack patterns and non-standard patterns such as
repeating binary patterns, keystroke patterns, downloaded daemons
or errors such as buffer overflow attempts. By observing the
operations performed on decoy operating system 32 the attack may be
categorized and analyzed to determine for example how an attack
gains control of decoy operating system 32. Any method of analyzing
the incoming data such as binary matching, neural network matching
or keyword matching or any other method of matching
attack-identifying information is in keeping with the spirit of the
invention.
[0032] In step 48, a decision is made as to whether to generate an
attack signature. If no harmful operations occurred as a result of
attack 10 or when no known attack patterns are found, then no
further attack inquiry would be needed as shown in step 50. The
processing module 40 can then take on the next input of captured
information from the monitor/intercept module 30.
[0033] If a determination is made that attack signature generation
is warranted, an attack signature is generated as illustrated in
step 52. Processing module 40 may generate a signature whenever
data is found to be transferred through the socket in one or more
embodiments of the invention. Alternatively, if the attack
signature already exists or if the data transfer is of a nature
that indicates probing rather than attack, then the attack
signature may not be generated. For example, processing module 40
may not generate a signature when it is found that no data has been
transferred through the socket even though the socket may have been
opened and closed without data transfer. Once the attack signature
is generated, the signature can be reviewed by the system
administrator who decides to send the attack signature, shown in
step 56, to the intrusion detection system (IDS) or intrusion
prevention system (IPS) for the protected network 20 through a
standard network connection including a wireless connection that is
generally not sent on protected network 20 or any other network
that the attacker may observe. This is accomplished by applying the
generated attack signature to the IDS/IPS library of signatures to
update the information contained in the library of signatures to
prevent the attacker from accessing the primary network with a
zero-day attack.
[0034] Embodiments of step 56 may save the generated attack
signatures in a database for future use or further analysis by
system administrators. The signatures may also be sent to a
proprietary global database of attack signatures for further
analysis. Any IDS/IPS may be utilized in one or more embodiments of
the invention. Existing IDS/IPS systems for example may be
interfaced with in order to integrate with existing solutions.
[0035] FIG. 3 illustrates a human readable summary of an example
attack. Line 300 shows that the file "msprexe.exe" is copied into
the "System" directory. Line 301 shows a first registry entry
created by the attack. Line 302 shows a second registry entry
created by the attack. Any other changes to the system may be
shown, as part of the attack-identifying information and the
information shown in FIG. 3 is exemplary only.
[0036] FIG. 4 illustrates an XML formatted attack signature
generated from the attack summarized in FIG. 3 for transmittal to
an IDS or IPS. XML block 400 includes tags that define the attack
signature in the format of the particular IDS or IPS. Any tags used
by any IDS or IPS are in keeping with the spirit of the invention
and the tags shown in FIG. 4 are exemplary only. For example any
ports, protocols, severity levels, alarm levels, signature name or
any other quantity may be utilized to inform an IDS or IPS of an
attack signature.
[0037] Another embodiment of a system for analyzing and preventing
unauthorized intrusion into a computer network is shown in FIG. 5.
This embodiment is directed to an improved method and system for
analyzing unauthorized intrusion into a decoy computer network, the
analysis of which is used to prevent unauthorized access into a
protected computer network. An embodiment of such a system is
illustrated in FIG. 5, while the method remains as shown in the
flowchart in FIG. 2 above.
[0038] The system 500, as shown in FIG. 5, includes a decoy
computer network 502 and a protected computer network 504, each
comprising one or more separate computing devices. The decoy
computer network 502 includes a virtualized operating system module
506 for monitoring the decoy network 502, and a processing module
508 for obtaining, analyzing, and responding to exploits.
[0039] These modules may be hosted on the same computing device or
on separate computing devices. However, for ease of explanation,
these modules will be described below as being hosted on separate
computing devices. Furthermore, although not shown, one skilled in
the art will appreciate that each of these computing devices may
include one or more processors, input/output devices, communication
circuitry, power sources, memory (both physical, e.g., RAM, and
disks, e.g., hard disk drives), and any other physical hardware
necessary for hosting and running the aforementioned modules. In
some embodiments, the modules 506 and 508 are as present in
physical memory once the system has been booted and is
operational.
[0040] The virtualized operating system module 506 includes a
hypervisor operating system 510 (also known as a virtual machine
monitor operating system) that provides a virtualization platform
that allows multiple virtual operating systems to be run on a host
computing device at the same time. In some embodiments, the
hypervisor operating system 510 is a LINUX-based system. One or
more fully-functioning "guest" virtualized operating systems 512
are run on the hypervisor operating system 510 at a level above the
hardware. As will be described in detail below, these virtualized
operating systems 512 act as decoy operating systems to attract
attacker activity 550. Any operating system may be used as guest
decoy operating system 512, including but not limited to WINDOWS,
SUN MICROSYSTEMS, SOLARIS, or any version of LINUX known to persons
skilled in the art, as well as any combination of the
aforementioned. It should be appreciated that all known operating
systems are within the scope of the present invention. There is
also no limit to either the number of virtualized guest decoy
operating systems 512 or the number of virtualized guest operating
system modules 506 that may be utilized.
[0041] Also running on the hypervisor operating system 510 are
normal hypervisor operating system userland processes 514. The
hypervisor operating system 510 includes a hypervisor kernel 516,
which in some embodiments is also Linux-based. The hypervisor
kernel 516 is that part of the hypervisor operating system 510 that
resides in physical memory at all times and provides the basic
services to the hypervisor operating system 510. The hypervisor
kernel 516 is the part of the operating system that activates the
hardware directly or interfaces with another software layer that,
in turn, drives the hardware. The virtualized decoy operating
systems 512 access the physical memory assigned to them by the
hypervisor operating system via the hypervisor kernel 516.
[0042] The hypervisor kernel 516 includes a hypervisor virtual
machine kernel module 518 that supports virtualization of the
"guest" decoy operating systems 512. The hypervisor kernel 516 also
includes virtual-machine-based rootkit module 520 coupled to the
hypervisor virtual machine kernel module 516. The
virtual-machine-based rootkit module 520 is a set of software tools
that conceal running processes, files or system data from the
virtualized decoy operating systems 512. As described in further
detail below, the virtual-machine-based rootkit module 520 is part
of introspection module 538, which performs introspection into the
physical memory segments assigned to each of the virtualized decoy
operating systems 512.
[0043] Virtual-machine-based rootkit userland processes 522 run on
top of the virtual-machine-based rootkit module 520. Together, the
rootkit module 520 and its associated userland processes 522
constitute the system's introspection module 538 (described further
below). Virtual-machine-based rootkit userland processes 522 also
pass data from the introspection module 538 to the processing
module 508.
[0044] In use, attacker activity 550 is directed at the decoy
computer network 502 through one or more ports of each of the
virtualized decoy operating systems 512 that are left open as a
gateway for attacker activity 550. For example, the decoy network
502 can be configured to respond to connection attempts made at
network addresses that do not exist on the protected network 504.
Connections to these non-existent network addresses are assumed to
be malicious, since no production hardware exists on the protected
network 504 at these addresses. Decoys 512 (in the form of a
virtualized operating system) may be configured to respond to any
such non-existent network address. As in a typical attack, the
attacker activity 550 scans for an open port, ostensibly in an
attempt to make a network connection and then access one or more
computing devices on the protected computer network 504. When the
attacker activity 550 scans for open ports at non-existent network
addresses, however, the attacker is presented with a virtualized
decoy operating system 512 instead.
[0045] When the attacker activity 550 connects to a virtualized
decoy operating system 512 through an open port, the attacker sees
a fully-functional standard operating system fingerprint. Since the
virtualized operating system module 506 can be configured to
present any operating system as a fully-functional virtualized
decoy 512, responses to connection requests from attacker activity
550 are guaranteed to be authentic for the operating system running
on that decoy. For example, an FTP port access request for WINDOWS
may return a specific character sequence that differs from an FTP
response for LINUX. Similarly, an FTP access request to a WINDOWS
port may return a response ">ftp: connect: Connection refused."
This character sequence may be slightly different from that
generated by LINUX. Further, different versions of WINDOWS may
respond with slightly different, version-specific character
sequences. Since attackers often use these sequences to identify
what type of operating system is at a particular network address
and the version (or range of possible versions) for that operating
system, the fact that virtualized decoy operating systems 512
generate authentic responses makes them realistic decoys and
encourages intruders to access them. The instigator of the attack
550 is thus lured into accessing the decoy 512, which is overseen
by the hypervisor operating system 510 running on the
hardware-based, virtualized operating system module 506. Attacker
activity 550 may then initiate custom or known exploits for the
observed operating system. When the attacker activity 550 proceeds
to interact with the decoy 512, the attacker provides the decoy 512
with the data used to obtain control of the decoy 512. These data
are recorded and analyzed without the knowledge of the attacker, as
described further below.
[0046] All scans by the attacker activity 550 receive real-world
operating system and service information, leading the instigator of
the attack 550 to believe that there is a potentially vulnerable
system responding. The attacker is thus lured into communicating
with virtualized operating system module 506 and its virtualized
decoy operating systems and services. Since real hardware is
utilized, the attacker is essentially attacking an actual physical
system and, therefore, cannot tell that the system is actually an
instrumented honeypot that monitors the attacker activity 550 from
the introspection module 538 described below.
[0047] As described above, the virtualized guest operating system
module 506 includes the virtual machine-based rootkit module 520
and its associated userland processes 522. Since both the virtual
machine-based rootkit module 520 and its associated userland
processes 522 run completely outside the virtualized decoy
operating systems 512, they remain hidden from the instigator of
the attack, with no discoverable impact on the decoy operating
systems' 512 performance. In one embodiment, the virtual
machine-based rootkit module 520 and its associated userland
processes 522 constitute an introspection module 538 (also known as
a virtual machine-based memory introspection analysis tool) that
monitors and introspects into the virtualized decoy operating
systems' memory segments. This occurs from within the hypervisor
operating system 510. The introspection module 538 introspects and
gathers information on any virtualized operating system supported
by the hypervisor operating system 510.
[0048] The introspection module 538 comprising the
virtual-machine-based rootkit module 520 and its associated
userland processes 522 examines the memory assigned to virtualized
decoy operating systems 512 in order to acquire low-level data
about the interaction between the decoy operating systems and
attack activity 500. The introspection module 538 examines the
memory of virtualized decoy operating systems 512 by means of three
functional components: a code region selector, a trace
instrumentor, and a trace analyzer. Regular expressions (also known
as `regex`) are used throughout the process to identify, describe,
and profile the contents of the virtualized decoy's memory
segments. The code selector identifies regions of code in memory
that are of interest for further introspection. Regions of interest
may include, but are not limited to, system calls, the arguments of
system calls, the returns of system calls, device and memory
input-output, driver information, library calls, branching
information, instruction pointer jumps, and raw network
information. The instrumentor copies the memory traces of interest
identified by the code selector and then profiles and instruments
them. The trace analyzer takes the instrumented traces and uses
them to replay the memory behavior of the decoy operating system
512. In this manner, the introspection module 538 examines the
contents of the decoy operating systems' 512 memory segments in an
instrumented context that generates and retrieves forensic data for
analysis by the processing module 508.
[0049] When an attacker connects to a network port and begins
interacting with a virtualized decoy operating system 512, the
introspection module 538 monitors and captures information from the
connection, including port numbers, data streams, file uploads,
keystrokes, ASCII or binary files, malicious payloads, memory
manipulation attempts, and any other data transfers or malicious
attempts.
[0050] The captured information, containing attack-identifying
information, is then sent from the introspection module 538 to the
processing module 508 by means of a virtual machine-based rootkit
userland process 522.
[0051] The processing module 508 includes an operating system
kernel 526, which in some embodiments is also LINUX based. The
processing module 508 also includes a database, such as a
relational database server 528, and a signature-generation engine
530. In some embodiments, the signature-generation engine 530
communicates with the introspection module 538 over a private
network interface communications channel 534 and accepts
custom-formatted protocol packets named BAT (Blade Activity
Transfer). The private network interface communications channel 524
may be a persistent Layer 3 TCP socket communications link that
cannot be seen or accessed by the attacker (such as an IP
connection or any other hardware-dependent custom communication
protocol known to persons skilled in the art). Thus, the processing
module 508 is part of a secure and separate administrative
network.
[0052] In use, the introspection module 538 captures (through
introspection) attack information. The attack information is then
communicated through the private network interface channel 524 and
stored on the relational database server 528 as forensic data for
later analysis. The signature-generation engine 530 then uses this
forensic data to generate a signature of the attack. The entire
process from attack detection through signature generation may
occur automatically, i.e., without any human intervention, at a
timescale ranging from nearly immediate to several minutes. The
intrusion prevention system (described below) uses the attack
signature to identify and prevent subsequent attacks.
[0053] The protected computer network 504 includes an IDS/IPS
library of signatures 534 and an IDS/IPS system 542 coupled to
multiple protected network devices 536. Suitable IDS/IPS systems
542 include Cisco Systems' IPS 4200 Series, Juniper's IDP 200, and
Enterasys' Dragon IDS Network Sensor.
[0054] In one or more embodiments, the signature may be sent from
the back-end processing module 508 to the intrusion detection
and/or prevention (IDS/IPS) signature library 534 through a second
network connection 540, which is used by the processing module 508
to directly interact with the IDS/IPS system 542. The
virtual-machine-based rootkit module 520 may easily clean the
virtualized decoy operating system 512 at any time by removing the
running system image of the compromised virtualized decoy operating
system and replacing it with a pre-attack system image. Thus the
virtual-machine-based rootkit module 520 can cleanse or reset the
virtualized decoy operating system of any malicious software or
payload, removing the possibility that attacker(s) can use that
virtualized decoy operating system 512 for further attacks on other
networks. In this manner, the attack can be thwarted, and the
operating system does not become a tool of the attacker(s). This
procedure may also be automated, i.e., may occur without further
human intervention.
[0055] As shown in FIG. 2, processing starts at Step 200 and waits
for activity from the introspection module 538 at Step 43. At Step
44, the processing module 508 generates a report of the attack that
includes attack-identifying information (See FIG. 3). This report
is for review and use by a system administrator responsible for the
security of a protected network 504. The attack may contain, but is
not limited to, one or more data transfers or keystrokes, which are
analyzed at Step 46. By observing whether the attacker is
successful in interacting with the system (i.e., if the system is
responding in a manner that shows that the attacker has gained
access), a determination can be made at Step 48 as to whether an
attack signature should be generated, and the attack signature is
created at step 52 (See FIG. 4). If the attacker, for example, is
unsuccessful at gaining access, or if there is no data transfer,
the attack inquiry may be ended at Step 50. Any attack signature
generated is sent to the IDS/IPS signature library 534 at Step 56,
and processing continues at Step 43.
[0056] In one embodiment of the invention, the report of the attack
is written and then displayed via a visualization interface 532 and
can include information about which sockets were accessed by the
attack 550, what happened at a particular socket, the keystrokes
entered or data transferred, what files were transferred, how the
attack 550 was run, what happened on the virtualized operating
system module 506, and how the virtualized decoy operating systems
512 running on the hypervisor operating system 510 and any related
network services were affected. In some embodiments, the
visualization interface 532 is AJAX- and/or FLASH-based. The report
may also include information on the location of the instigator of
the attack 550 or the service provider used for the attack.
Graphical representations of key information and interactive
mapping of attack locales by region or country may also be included
in the report. The visualization interface may also be used to
analyze, configure, and automate the system's response to attack
activity 550 on timescales ranging from near-immediate to several
minutes from the initiation of an attack.
[0057] At Step 46, the attack-identifying information is analyzed
for known attack patterns as well as non-standard patterns, such as
repeating binary patterns, keystroke patterns, downloaded daemons,
or errors (such as buffer overflow attempts, malicious payloads
attempting to execute arbitrary code on the system, memory
overwriting attempts, stack attacks, and heap attacks). By
observing the operations performed on the decoy operating system(s)
512, the attack 550 may be categorized and analyzed to determine,
for example, how an attack gained control of the decoy operating
system(s) 512. Any method of analyzing the incoming data such as
binary matching, neural-network matching, keyword matching, or any
other method of matching attack-identifying information is in
keeping with the spirit of the invention. Pattern-matching
techniques involving neural networks, for example, are
characterized in Carl Looney's Pattern Recognition Using Neural
Networks. Theory and Algorithms for Engineers and Scientists
(Oxford University Press USA, New York, N.Y., 1997) and Christopher
Bishop's Neural Networks for Pattern Recognition (Oxford University
Press USA, New York, N.Y., 1995), among other sources familiar to
those skilled in the art.
[0058] At Step 48, a decision is made as to whether to generate an
attack signature. If no harmful operations occurred as a result of
an attack, or when no known attack patterns are found, then no
further attack inquiry would be needed (as shown at Step 50). The
processing module 508 may then take on the next input of captured
information from the introspection module 538 running on the
hardware-based, virtualized operating system module 506.
[0059] If a determination is made that attack signature generation
is warranted, an attack signature is generated as illustrated in
Step 52. In one or more embodiments of the invention, the
processing module 508 may generate a signature whenever data is
found to be transferred through the socket. Alternatively, if the
attack signature already exists, or if the data transfer is of a
nature that indicates probing rather than attack, then the attack
signature may not be generated. For example, the processing module
508 may not generate a signature when it is found that no data has
been transferred through the socket, even though the socket may
have been opened and closed. The conditions under which the
processing module 508 generates an attack can be configured and
automated by an administrator. Once the attack signature is
generated, the signature can be reviewed by the system
administrator, who decides whether to send the attack signature
(shown at Step 56) to the intrusion detection system (IDS) or
intrusion prevention system (IPS) for the protected network 504.
The attack signature is sent through a standard network connection
or via a wireless connection and is generally sent on a private
portion of the protected network 504 that the attacker cannot
observe. The generated attack signature is thus applied to the
IDS/IPS library of signatures 534, thereby updating the information
contained in the signature library and preventing the attacker from
accessing the protected network 504.
[0060] Embodiments of the invention may save the attack signatures
created at Step 52 in a relational database server 528 for future
use or analysis by system administrators. The signatures may also
be sent to a proprietary global database of attack signatures for
further analysis, storage, and distribution. Any IDS/IPS system may
be utilized in one or more embodiments of the invention. The
invention may be interfaced with existing IDS/IPS systems, for
example to integrate it with existing solutions.
[0061] As explained above, FIG. 3 illustrates a human-readable
summary of an example attack. Line 300 shows that the file
"msprexe.exe" is copied into the "System" directory. Line 301 shows
a first registry entry created by the attack. Line 302 shows a
second registry entry created by the attack. Any other changes to
the system may be shown as part of the attack-identifying
information, and the information shown in FIG. 3 is exemplary
only.
[0062] As explained above, FIG. 4 illustrates an attack signature
generated from the attack summarized in FIG. 3 and formatted in XML
for transmission to an IDS or IPS. XML Block 400 includes tags that
define the attack signature in the format of the particular IDS or
IPS. Any tags used by any IDS or IPS are in keeping with the spirit
of the invention, and the tags shown in FIG. 4 are exemplary only.
For example, any ports, protocols, severity levels, alarm levels,
signature name, or any other quantity, may be utilized to inform an
IDS or IPS of an attack signature.
[0063] While embodiments and alternatives have been disclosed and
discussed, the invention herein is not limited to the particular
disclosed embodiments or alternatives but encompasses the full
breadth and scope of the invention including equivalents, and the
invention is not limited except as set forth in and encompassed by
the full breadth and scope of the claims herein.
* * * * *