U.S. patent application number 13/293705 was filed with the patent office on 2012-11-15 for coarse bin frequency synchronization in a communication system.
This patent application is currently assigned to QUALCOMM INCORPORATED. Invention is credited to Seguei A. GLAZKO, Kuei-Chiang LAI, Shimman PATEL.
Application Number | 20120288037 13/293705 |
Document ID | / |
Family ID | 38173422 |
Filed Date | 2012-11-15 |
United States Patent
Application |
20120288037 |
Kind Code |
A1 |
PATEL; Shimman ; et
al. |
November 15, 2012 |
COARSE BIN FREQUENCY SYNCHRONIZATION IN A COMMUNICATION SYSTEM
Abstract
For frequency bin error estimation, multiple hypotheses are
formed for different frequency bin errors, pilot offsets, or
combinations of frequency bin error and pilot offset. For each
hypothesis, received symbols are extracted from the proper subbands
determined by the hypothesis. In one scheme, the extracted received
symbols for each hypothesis are despread with a scrambling sequence
to obtain despread symbols for that hypothesis. A metric is derived
for each hypothesis based on the despread symbols, e.g., by
deriving a channel impulse response estimate based on the despread
symbols and then deriving the metric based on the channel impulse
response estimate. In another scheme, the extracted received
symbols for each hypothesis are correlated, and a metric is derived
based on the correlation results. For both schemes, the frequency
bin error and/or the pilot offset are determined based on the
metrics for all hypotheses evaluated.
Inventors: |
PATEL; Shimman; (San Diego,
CA) ; LAI; Kuei-Chiang; (Chutung, TW) ;
GLAZKO; Seguei A.; (San Diego, CA) |
Assignee: |
QUALCOMM INCORPORATED
|
Family ID: |
38173422 |
Appl. No.: |
13/293705 |
Filed: |
November 10, 2011 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
11313555 |
Dec 20, 2005 |
8130726 |
|
|
13293705 |
|
|
|
|
Current U.S.
Class: |
375/340 |
Current CPC
Class: |
H04L 25/022 20130101;
H04L 27/2659 20130101; H04L 25/0228 20130101; H04L 27/2613
20130101; H04L 27/2695 20130101; H04L 2027/0034 20130101; H04L
5/0048 20130101; H04L 27/3872 20130101; H04L 27/2662 20130101; H04L
2027/0065 20130101 |
Class at
Publication: |
375/340 |
International
Class: |
H04L 27/28 20060101
H04L027/28 |
Claims
1. An apparatus comprising: a wireless receiver, wherein the
wireless receiver comprises: a processor configured to perform
despreading of received symbols with a scrambling sequence for each
of a plurality of hypotheses, to derive a metric for each
hypothesis based on despread symbols for the hypothesis, and to
determine a frequency error based on metrics derived for the
plurality of hypotheses; and a memory coupled to the processor;
wherein the processor is configured to derive a channel impulse
response estimate for each hypothesis based on the despread symbols
for the hypothesis, and to derive the metric for each hypothesis
based on the channel impulse response estimate for the
hypothesis.
2. The apparatus of claim 1, wherein the processor is configured to
form the plurality of hypotheses for a range of frequency errors,
wherein each hypothesis corresponds to a different hypothesized
frequency error.
3. The apparatus of claim 1, wherein the processor is configured to
form the plurality of hypotheses for a range of frequency errors
and for multiple pilot offsets, wherein each hypothesis corresponds
to a different combination of frequency error and pilot offset.
4. The apparatus of claim 1, wherein for each hypothesis the
processor is configured to extract the received symbols for
subbands determined by the hypothesis, and to perform despreading
of the extracted received symbols with the scrambling sequence.
5. The apparatus of claim 4, wherein the extracted received symbols
are hypothesized to be for a scattered pilot sent on different sets
of subbands in different symbol periods.
6. The apparatus of claim 4, wherein the extracted received symbols
are hypothesized to be for a continual pilot sent on a
predetermined set of subbands.
7. The apparatus of claim 1, wherein the processor is configured to
derive the metric for each hypothesis based on energy of a largest
channel tap in the channel impulse response estimate for the
hypothesis.
8. The apparatus of claim 1, wherein the processor is configured to
identify large channel taps in the channel impulse response
estimate for each hypothesis based on a threshold, and to derive
the metric for each hypothesis based on energy of the large channel
taps for the hypothesis.
9. The apparatus of claim 1, wherein the received symbols are for
data and pilot transmitted using orthogonal frequency division
multiplexing (OFDM).
10. The apparatus of claim 1, wherein the received symbols are for
data and pilot transmitted using single-carrier frequency division
multiple access (SC-FDMA).
11. A method comprising: performing despreading of received symbols
with a scrambling sequence for each of a plurality of hypotheses;
deriving a metric for each hypothesis based on despread symbols for
the hypothesis; and determining a frequency error based on metrics
derived for the plurality of hypotheses. wherein the deriving the
metric for each hypothesis comprises: deriving a channel impulse
response estimate for each hypothesis based on the despread symbols
for the hypothesis; and deriving the metric for each hypothesis
based on the channel impulse response estimate for the
hypothesis.
12. The method of claim 11, further comprising: forming the
plurality of hypotheses for a range of frequency errors and for
multiple pilot offsets, wherein each hypothesis corresponds to a
different combination of frequency error and pilot offset.
13. The method of claim 11, further comprising: forming the
plurality of hypotheses for a range of frequency errors, wherein
each hypothesis corresponds to a different hypothesized frequency
error.
14. The method of claim 11, further comprising: extracting the
received symbols for subbands determined by the hypothesis; and
performing despreading of the extracted received symbols with the
scrambling sequence.
15. An apparatus comprising: means for performing despreading of
received symbols with a scrambling sequence for each of a plurality
of hypotheses; means for deriving a metric for each hypothesis
based on despread symbols for the hypothesis; and means for
determining a frequency error based on metrics derived for the
plurality of hypotheses. wherein the means for deriving the metric
for each hypothesis comprises: means for deriving a channel impulse
response estimate for each hypothesis based on the despread symbols
for the hypothesis; and means for deriving the metric for each
hypothesis based on the channel impulse response estimate for the
hypothesis.
16. The apparatus of claim 15, further comprising: means for
forming the plurality of hypotheses for a range of frequency errors
and for multiple pilot offsets, wherein each hypothesis corresponds
to a different combination of frequency error and pilot offset.
17. The apparatus of claim 15, further comprising: means for
forming the plurality of hypotheses for a range of frequency
errors, wherein each hypothesis corresponds to a different
hypothesized frequency error.
18. The apparatus of claim 15, further comprising: means for
extracting the received symbols for subbands determined by the
hypothesis; and means for performing despreading of the extracted
received symbols with the scrambling sequence.
19. A non-transitory computer readable medium containing software
that, when executed, causes the computer to perform the acts of:
performing despreading of received symbols with a scrambling
sequence for each of a plurality of hypotheses; deriving a metric
for each hypothesis based on despread symbols for the hypothesis;
and determining a frequency error based on metrics derived for the
plurality of hypotheses. wherein the deriving the metric for each
hypothesis comprises: deriving a channel impulse response estimate
for each hypothesis based on the despread symbols for the
hypothesis; and deriving the metric for each hypothesis based on
the channel impulse response estimate for the hypothesis.
Description
CLAIM OF PRIORITY UNDER 35 U.S.C. .sctn.120
[0001] The present application for patent is a divisional of patent
application Ser. No. 11/313,555 entitled "COARSE BIN FREQUENCY
SYNCHRONIZATION IN A COMMUNICATION SYSTEM" filed Dec. 20, 2008,
pending, and assigned to the assignee hereof and hereby expressly
incorporated by reference herein.
BACKGROUND
[0002] I. Field
[0003] The present disclosure relates generally to communication,
and more specifically to techniques for performing frequency
synchronization in a communication system.
[0004] II. Background
[0005] Orthogonal frequency division multiplexing (OFDM) is a
multi-carrier modulation technique that can provide good
performance for some wireless environments. OFDM partitions the
overall system bandwidth into multiple (K) orthogonal frequency
subbands, which are also called carriers, subcarriers, tones, and
so on. With OFDM, each subband is associated with a respective
carrier that may be modulated with data. In the following
description, "subband" and "carrier" are synonymous terms and are
used interchangeably.
[0006] In an OFDM system, a transmitter processes (e.g., encodes,
interleaves, and modulates) traffic data to generate modulation
symbols and further maps the modulation symbols to the K total
subbands. The transmitter then transforms the modulation symbols
for each OFDM symbol period to the time domain and forms an OFDM
symbol. The transmitter transmits the OFDM symbols to a
receiver.
[0007] The receiver performs the complementary processing on the
OFDM symbols received from the transmitter. The receiver transforms
each received OFDM symbol to the frequency domain to obtain K
received symbols for the K subbands. The received symbols are noisy
and distorted versions of the modulation symbols sent by the
transmitter. The receiver typically performs frequency
synchronization to determine frequency error at the receiver. The
frequency error may be due to difference in the oscillator
frequencies at the transmitter and the receiver, Doppler shift, and
so on. Frequency synchronization is challenging in certain channel
environments such as low signal-to-noise ratio (SNR) conditions,
fast fading, and so on. Furthermore, it is desirable to perform
frequency synchronization quickly so that the processing overhead
is as low as possible.
[0008] There is therefore a need in the art for techniques to
perform frequency synchronization in a communication system.
SUMMARY
[0009] Techniques for performing frequency synchronization in a
communication system are described herein. The frequency error at a
receiver may be decomposed into a fractional portion and an integer
portion. The fractional portion is less than one bin and may be
estimated and removed in a manner known in the art. A bin is the
spacing between adjacent subbands. The integer portion is also
called frequency bin error and is an integer number of bins. The
frequency bin error may be estimated using the techniques described
herein.
[0010] In an embodiment of frequency bin error estimation, multiple
hypotheses are initially formed for different frequency bin errors,
different pilot offsets, or different combinations of frequency bin
error and pilot offset. A pilot may be sent on different sets of
subbands, and each pilot offset corresponds to a different set of
subbands on which the pilot may have been sent. For each
hypothesis, received symbols are extracted from the proper subbands
determined by the hypothesis. The extracted received symbols are
hypothesized to be for (1) a scattered pilot that is sent on
different sets of subbands in different symbol periods and/or (2) a
continual pilot that is sent on a fixed set of subbands in all
symbol periods.
[0011] In an embodiment, the extracted received symbols for each
hypothesis are despread with a scrambling sequence to obtain
despread symbols for that hypothesis. The scrambling sequence is
used to generate the scattered and continual pilots at the
transmitter. A metric is then derived for each hypothesis based on
the despread symbols for that hypothesis, e.g., by deriving a
channel impulse response estimate based on the despread symbols and
then deriving the metric based on the channel impulse response
estimate. In another embodiment, the extracted received symbols for
each hypothesis are correlated, and a metric is derived for the
hypothesis based on the correlation results. For both embodiments,
the frequency bin error and/or the pilot offset are determined
based on the metrics for all hypotheses evaluated.
[0012] The frequency bin error estimation may also be performed in
other manners, as described below. Various aspects and embodiments
of the invention are described in further detail below.
BRIEF DESCRIPTION OF THE DRAWINGS
[0013] The features and nature of the present invention will become
more apparent from the detailed description set forth below when
taken in conjunction with the drawings in which like reference
characters identify correspondingly throughout.
[0014] FIG. 1 shows a block diagram of a transmitter and a
receiver.
[0015] FIG. 2 shows a subband structure.
[0016] FIGS. 3A and 3B show pilot structures for DVB-H and ISDB-T,
respectively.
[0017] FIG. 4 shows an OFDM demodulator at the receiver.
[0018] FIGS. 5 and 6 show two embodiments of a frequency bin error
estimator.
[0019] FIG. 7 shows a process for performing frequency error
estimation by despreading the received symbols.
[0020] FIG. 8 shows a process for performing frequency error
estimation by correlating the received symbols.
[0021] FIG. 9 shows a process for performing frequency error
estimation in stages.
DETAILED DESCRIPTION
[0022] The word "exemplary" is used herein to mean "serving as an
example, instance, or illustration." Any embodiment or design
described herein as "exemplary" is not necessarily to be construed
as preferred or advantageous over other embodiments or designs.
[0023] The frequency synchronization techniques described herein
may be used for various communication systems such as an OFDM
system, an orthogonal frequency division multiple access (OFDMA)
system, a single-carrier frequency division multiple access
(SC-FDMA) system, and so on. An OFDMA system utilizes OFDM. An
SC-FDMA system may utilize interleaved FDMA (IFDMA) to transmit on
subbands that are distributed across the system bandwidth,
localized FDMA (LFDMA) to transmit on a block of adjacent subbands,
or enhanced FDMA (EFDMA) to transmit on multiple blocks of adjacent
subbands. In general, modulation symbols are sent in the frequency
domain with OFDM and in the time domain with SC-FDMA.
[0024] For clarity, the techniques are specifically described below
for two exemplary OFDM-based systems that implement Digital Video
Broadcasting for Handhelds (DVB-H) and Integrated Services Digital
Broadcasting for Terrestrial Television Broadcasting (ISDB-T).
DVB-H and ISDB-T support digital transmission of multimedia over a
terrestrial communication network. DVB-H has 3 modes of operation
for FFT sizes of 2K, 4K and 8K. ISDB-T has 3 modes of operation for
FFT sizes of 256, 512 and 1K. DVB-H is described in document ETSI
EN 300 744, entitled "Digital Video Broadcasting (DVB); Framing
structure, channel coding and modulation for digital terrestrial
television," November 2004. ISDB-T is described in document ARIB
STD-B31, entitled "Transmission System for Digital Terrestrial
Television Broadcasting," July 2003. These documents are publicly
available.
[0025] FIG. 1 shows a block diagram of a transmitter 110 and a
receiver 150 in an OFDM-based system 100, which may implement
DVB-H, ISDB-T, and/or some other design. At transmitter 110, a
transmit (TX) data processor 120 receives and processes (e.g.,
formats, encodes, interleaves, and symbol maps) traffic data to
generate data symbols. As used herein, a data symbol is a
modulation symbol for traffic data, a pilot symbol is a modulation
symbol for pilot, which is data that is known a priori by both the
transmitter and receiver, and a zero symbol is a signal value of
zero.
[0026] An OFDM modulator 130 receives and multiplexes the data
symbols and pilot symbols onto data subbands and pilot subbands,
respectively. A data subband is a subband used to send traffic
data, and a pilot subband is a subband used to send pilot. A given
subband may serve as a data subband in one OFDM symbol period and
as a pilot subband in another OFDM symbol period. An OFDM symbol
period is the duration of one OFDM symbol and is also referred to
as a symbol period. The pilot symbols may be multiplexed with the
data symbols as described below. OFDM modulator 130 obtains K
transmit symbols for the K total subbands in each OFDM symbol
period. Each transmit symbol may be a data symbol, a pilot symbol,
or a zero symbol. OFDM modulator 130 transforms the K transmit
symbols for each OFDM symbol period with a K-point inverse fast
Fourier transform (IFFT) or inverse discrete Fourier transform
(IDFT) to obtain a transformed symbol that contains K time-domain
chips. OFDM modulator 130 then repeats a portion of the transformed
symbol to generate an OFDM symbol. The repeated portion is often
called a cyclic prefix or a guard interval and is used to combat
frequency selective fading, which is a frequency response that
varies across the system bandwidth due to multipath in a wireless
channel. OFDM modulator 130 provides an OFDM symbol for each OFDM
symbol period. A transmitter unit (TMTR) 132 receives and processes
(e.g., converts to analog, amplifies, filters, and frequency
upconverts) the OFDM symbols and generates a modulated signal,
which is transmitted via an antenna 134 to receiver 150.
[0027] At receiver 150, an antenna 152 receives the modulated
signal from transmitter 110 and provides a received signal to a
receiver unit (RCVR) 154. Receiver unit 154 conditions (e.g.,
filters, amplifies, frequency downconverts, and digitizes) the
received signal to obtain input samples. An OFDM demodulator
(Demod) 160 processes the input samples as described below and
obtains K received symbols for the K total subbands in each OFDM
symbol period. The received symbols include received data symbols
for the data subbands and received pilot symbols for the pilot
subbands. OFDM demodulator 160 performs frequency synchronization
to estimate and remove the frequency error at receiver 150. OFDM
demodulator 160 also performs data demodulation/detection on the
received data symbols with a channel estimate to obtain data symbol
estimates, which are estimates of the data symbols sent by
transmitter 110. A receive (RX) data processor 170 then processes
(e.g., symbol demaps, deinterleaves, and decodes) the data symbol
estimates to obtain decoded data. In general, the processing by
OFDM demodulator 160 and RX data processor 170 is complementary to
the processing by OFDM modulator 130 and TX data processor 120,
respectively, at transmitter 110.
[0028] Controllers/processors 140 and 180 control the operation of
various processing units at transmitter 110 and receiver 150,
respectively. Memories 142 and 182 store data and program codes for
transmitter 110 and receiver 150, respectively.
[0029] FIG. 2 shows an exemplary subband structure 200 for system
100. The overall system bandwidth of BW MHz is partitioned into
multiple (K) subbands that are given indices of 0 through K-1,
where K may be a configurable value. The spacing between adjacent
subbands is BW/K MHz. For subband structure 200, the K total
subbands are arranged into 12 disjoint interlaces. The 12
interlaces are disjoint in that each of the K subbands belongs in
only one interlace. Each interlace contains approximately K/12
subbands that are uniformly distributed across the K total subbands
such that consecutive subbands in the interlace are spaced apart by
12 subbands. Thus, interlace u, for u.epsilon.{0, . . . , 11},
contains subbands u, u+12, u+24, . . . . Index u is the interlace
index as well as a subband offset that indicates the first subband
in the interlace. FIG. 2 only shows four interlaces 0, 3, 6 and
9.
[0030] FIG. 3A shows a pilot structure 300 for DVB-H. Pilot
structure 300 includes a continual pilot and a scattered pilot. The
continual pilot is sent on C subbands that are distributed across
the system bandwidth, where C is dependent on the mode. The pilot
is continual in that it is sent on the same C subbands in all OFDM
symbol periods. These C subbands include subbands 0, 48, 54, . . .
, K-1 and are given in ETSI EN 300 744. The scattered pilot is sent
on one interlace in each OFDM symbol period. The transmission
timeline for DVB-H is partitioned into frames, with each frame
including 68 OFDM symbols that are given indices of 0 through 67.
The scattered pilot is sent on interlace 0 in OFDM symbol 0,
interlace 3 in OFDM symbol 1, interlace 6 in OFDM symbol 2,
interlace 9 in OFDM symbol 4, interlace 0 in OFDM symbol 5, and so
on. The scattered pilot is thus sent on the same four interlaces in
each set of 4 OFDM symbols.
[0031] FIG. 3B shows a pilot structure 310 for ISDB-T. Pilot
structure 310 includes only a scattered pilot that is sent on
interlaces 0, 3, 6 and 9 in each set of 4 OFDM symbols. The
transmission timeline for ISDB-T is also partitioned into frames,
with each frame including 204 OFDM symbols that are given indices
of 0 through 203. The scattered pilot is sent on interlace 0 in
OFDM symbol 0 and cycles through interlaces 0, 3, 6 and 9 in the
same manner as the scattered pilot for DVB-H.
[0032] For both DVB-H and ISDB-T, the pilot symbols for each OFDM
symbol are generated based on a pseudo-random binary sequence
(PBRS) that is derived from a specific generator polynomial. The
PBRS sequence contains K bits and is given as:
{w}={w.sub.0,w.sub.1,w.sub.2,w.sub.3,w.sub.4, . . . ,w.sub.K-1}. Eq
(1)
PBRS bit W.sub.k, for k.epsilon.{0, . . . , K-1}, is used to
generate a BPSK modulation symbol that is used as a pilot symbol
for subband k. The pilot symbols for interlace u, for u.epsilon.{0,
3, 6, 9}, are generated with PBRS bits {w.sub.u, w.sub.u+12,
w.sub.u+24, w.sub.u+36, . . . }.
[0033] Table 1 lists the values for some parameters for the three
modes in DVB-H and ISDB-T. In Table 1, parameters N, K, C and S are
given for one OFDM symbol. The number of scattered pilot subbands
(S) for both DVB-H and ISDB-T and the number of continual pilot
subbands (C) for DVB-H are dependent on the mode. For ISDB-T, K is
an integer multiple of 12, and interlaces 0, 3, 6 and 9 contain the
same number of pilot subbands. For DVB-H, K is not an integer
multiple of 12, and interlace 0 contains one more pilot subband
than interlaces 3, 6 and 9. For simplicity, the following
description assumes that the interlaces contain the same number of
(S) pilot subbands.
TABLE-US-00001 TABLE 1 DVB-H ISDB-T Nota- Mode Description tion 1 2
3 1 2 3 FFT size N 2048 4096 8192 256 512 1024 Total number of K
1705 3409 6817 108 216 432 subbands Number of continual C 45 89 177
-- -- -- pilot subbands Number of scattered S 142 284 568 9 18 36
pilot subbands
In Table 1, the FFT size is more than twice the total number of
subbands for ISDB-T in order to relax the front-end filtering
requirements while still maintain a low level of aliasing
noise.
[0034] FIG. 4 shows a block diagram of an embodiment of OFDM
demodulator 160 at receiver 150 in FIG. 1. Within OFDM demodulator
160, a pre-processor 410 receives and processes the input samples
from receiver unit 154 and provides pre-processed samples.
Pre-processor 410 may perform automatic gain control (AGC), timing
acquisition, filtering, sample rate conversion, direct current (DC)
offset removal, and/or other functions. The pre-processed samples
have a frequency error that may be expressed as:
f.sub.err=mf.sub.bin+.DELTA.f, Eq (2)
where
[0035] f.sub.err is the total frequency error at the receiver;
[0036] .DELTA.f is the fractional portion of the frequency error,
which is less than one bin;
[0037] f.sub.bin is one bin, which is the spacing between adjacent
subbands; and
[0038] m is the integer portion of the frequency error, which is an
integer number of bins.
The integer portion of the frequency error is also called frequency
bin error or coarse bin frequency error.
[0039] A coarse frequency estimator 412 estimates the fractional
frequency error .DELTA.f based on the pre-processed samples and in
a manner known in the art. A rotator 414 receives the estimated
fractional frequency error .DELTA.{circumflex over (f)} from
estimator 412 and the estimated frequency bin error {circumflex
over (m)} from a frequency bin error estimator 420, removes the
estimated total frequency error from the pre-processed samples, and
provides frequency-corrected samples. A cyclic prefix removal unit
416 removes the cyclic prefix appended to each OFDM symbol and
provides received samples.
[0040] An FFT/DFT unit 418 performs a fast Fourier transform (FFT)
or discrete Fourier transform (DFT) on the received samples for
each OFDM symbol period and provides frequency-domain received
symbols for the K total subbands. Frequency bin error estimator 420
estimates the frequency bin error based on the received pilot
symbols and provides the estimated frequency bin error, as
described below. Rotator 414 may remove the estimated frequency bin
error from the pre-processed samples, as shown in FIG. 4.
Alternatively, a frequency bin correction unit can remove the
estimated frequency bin error from the received data symbols (not
shown in FIG. 4). A channel estimator 422 derives a channel
estimate based on the received pilot symbols. The channel estimate
may be a time-domain channel impulse response estimate or a
frequency-domain channel frequency response estimate. A data
demodulator 424 performs data demodulation/detection on the
received data symbols with the channel estimate and provides data
symbol estimates.
[0041] Although not shown in FIG. 4 for simplicity, OFDM
demodulator 160 may include processing units for fine frequency
tracking, fine time tracking, frame synchronization, and/or other
functions.
[0042] Frequency bin error estimator 420 estimates the frequency
bin error and further determines the scattered pilot offset, which
indicates the specific interlace used for the scattered pilot in
each OFDM symbol period. The maximum frequency bin error is
determined by the accuracy of the reference oscillator at receiver
150, the center frequency of the modulated signal being received,
and the mode used by the system. For example, if the reference
oscillator has a maximum error of 5 parts per million (ppm) and the
center frequency is 800 MHz, then the maximum frequency error is
.+-.4 KHz. This .+-.4 KHz frequency error corresponds to .+-.4 bins
for a subband spacing of 1116 Hz for mode 3 in ISDB-T and to .+-.6
bins for a subband spacing of 697 Hz for mode 3 in DVB-H. For
ISDB-T, there is an ambiguity of .+-.4 bins. Hence, the correct
frequency bin error is one of 9 "frequency" hypotheses for -4, -3,
-2, -1, 0, +1, +2, +3 and +4 bin errors.
[0043] Receiver 150 typically does not have frame timing when first
tuned to transmitter 110. In this case, for a given OFDM symbol,
receiver 150 does not know whether the scattered pilot is being
sent on interlace 0, 3, 6 or 9. As shown in FIG. 2, a pilot offset
of 0 corresponds to the scattered pilot being sent on interlace 0,
a pilot offset of 1 corresponds to the scattered pilot being sent
on interlace 3, a pilot offset of 2 corresponds to the scattered
pilot being sent on interlace 6, and a pilot offset of 3
corresponds to the scattered pilot being sent on interlace 9. There
is thus an ambiguity of 4 pilot offsets. Hence, the correct pilot
offset is one of 4 "time" hypotheses for pilot offsets of 0, 1, 2
and 3.
[0044] The frequency bin error estimation may be performed in
various manners. In an embodiment, the estimation is performed
based on an assumption that both frequency bin error and pilot
offset are unknown. For this embodiment, multiple hypotheses are
formed jointly for frequency and time. In another embodiment, the
estimation is performed in two steps, with the first step
determining the frequency bin error and the second step determining
the pilot offset. For this embodiment, multiple hypotheses are
formed separately for frequency and time. The frequency bin error
estimation may also be performed based on various metrics. In an
embodiment, the estimation is performed based on metrics derived
from despreading the received symbols. In another embodiment, the
estimation is performed based on metrics derived from correlating
the received symbols.
[0045] Table 2 lists four exemplary frequency bin error estimation
schemes, the hypotheses and metrics for each scheme, and the
system(s) for which each scheme is applicable. For clarity, schemes
1 and 4 are specifically described below.
TABLE-US-00002 TABLE 2 Scheme Hypotheses Metrics System(s) 1 Joint
frequency Despreading-based DVB-H and and time ISDB-T 2 Joint
frequency Correlation-based DVB-H and and time ISDB-T 3 Separate
frequency Despreading-based DVB-H and time 4 Separate frequency
Correlation-based DVB-H and time
[0046] For frequency bin error estimation scheme 1 in Table 2,
multiple frequency/time hypotheses are formed for different
combinations of frequency bin error and pilot offset. The total
number of frequency/time hypotheses to evaluate is equal to the
product of the number of hypotheses for frequency bin error (for
frequency uncertainty) and the number of hypotheses for pilot
offset (for time uncertainty), which is 9.times.4=36 frequency/time
hypotheses for the example described above for ISDB-T. One
frequency/time hypothesis is the correct hypothesis for both
frequency bin error and pilot offset, and the remaining
frequency/time hypotheses are incorrect.
[0047] The received symbols at receiver 150, without any frequency
error, may be expressed as:
Z.sub.k(l)=H.sub.k(l)S.sub.k(l)+N.sub.k(l), Eq (3)
where
[0048] S.sub.k(l) is a modulation symbol sent on subband k in OFDM
symbol period 1;
[0049] H.sub.k(l) is a channel gain for subband k in OFDM symbol
period 1;
[0050] Z.sub.k(l) is a received symbol for subband k in OFDM symbol
period 1; and
[0051] N.sub.k(l) is the noise for subband k in OFDM symbol period
1.
S.sub.k(l) may be a data symbol or a pilot symbol. The pilot
symbols are generated based on the PBRS sequence, and the pilot
symbol for subband k may be given as S.sub.k(l)=(4/3)w.sub.k, where
4/3 is a scaling factor for pilot relative to data.
[0052] If the frequency error is x bins, and assuming that the
fraction frequency error .DELTA.f has been removed by rotator 414,
then the received symbols for OFDM symbol periods l and l+1 may be
expressed as:
Z.sub.k+x(l)=H.sub.k(l)S.sub.k(l)+N.sub.k+x(l), and Eq (4)
Z.sub.k+x(l+1)=e.sup.j2.pi.xGH.sub.k(l+1)S.sub.k(l+1)+N.sub.k+x(l+1),
Eq (5)
where G is a guard interval ratio. As shown in equations (4) and
(5), a frequency error of x bins results in the modulation symbol
sent on subband k being received on subband k+x at the receiver.
The factor e.sup.j2.pi.xG is due to phase rotation in the received
symbols for OFDM symbol l+1 relative to the phase of the received
symbols for OFDM symbol l with a frequency error of x bins.
[0053] In an embodiment, each frequency/time hypothesis covers a
set of four consecutive OFDM symbols l through l+3. A given
frequency/time hypothesis H.sub.x,y, which corresponds to a
hypothesized frequency error of x bins and a hypothesized pilot
offset of y, may be evaluated as follows. First, the received
symbols are extracted from pilot subbands corresponding to
frequency bin error x and pilot offset y. In particular, received
symbols are extracted from interlaces x, x+3, x+6 and x+9 in the
four OFDM symbols for y=0, from interlaces x+3, x+6, x+9 and x in
the four OFDM symbols for y=1, from interlaces x+6, x+9, x and x+3
in the four OFDM symbols for y=2, and from interlaces x+9, x, x+3
and x+6 in the four OFDM symbols for y=3. The extracted received
symbols for each OFDM symbol are then despread with the
corresponding bits of the PBRS sequence to obtain despread symbols.
The despread symbols for OFDM symbols l+1, l+2 and l+3 are
multiplied with e.sup.-j2.pi.xG, e.sup.-j4.pi.xG and
e.sup.-j6.pi.xG, respectively, to account for the phase rotation
across OFDM symbols due to the frequency error of x bins. The
results of the processing are estimated channel gains (or simply,
channel gains) for the pilot subbands. The channel gains for
hypothesis H.sub.x,y for pilot offsets of y=0, 1, 2 and 3 are given
below:
H x , 0 = { [ w 0 Z x ( l ) , w 12 Z x + 12 ( l ) , w 24 Z x + 24 (
l ) , , w T Z x + T ( l ) ] [ w 3 Z x + 3 ( l + 1 ) , w 15 Z x + 15
( l + 1 ) , w 27 Z x + 27 ( l + 1 ) , , w T + 3 Z x + T + 3 ( l + 1
) ] - j2.pi. x G [ w 6 Z x + 6 ( l + 2 ) , w 18 Z x + 18 ( l + 2 )
, w 30 Z x + 30 ( l + 2 ) , , w T + 6 Z x + T + 6 ( l + 2 ) ] - j 4
.pi. x G [ w 9 Z x + 9 ( l + 3 ) , w 21 Z x + 21 ( l + 3 ) , w 33 Z
x + 33 ( l + 3 ) , , w T + 9 Z x + T + 9 ( l + 3 ) ] - j6.pi. x G }
, H x , 1 = { [ w 3 Z x + 3 ( l ) , w 15 Z x + 15 ( l ) , w 27 Z x
+ 27 ( l ) , , w T + 3 Z x + T + 3 ( l ) ] [ w 6 Z x + 6 ( l + 1 )
, w 18 Z x + 18 ( l + 1 ) , w 30 Z x + 30 ( l + 1 ) , , w T + 6 Z x
+ T + 6 ( l + 1 ) ] - j2.pi. x G [ w 9 Z x + 9 ( l + 2 ) , w 21 Z x
+ 21 ( l + 2 ) , w 33 Z x + 33 ( l + 2 ) , , w T + 9 Z x + T + 9 (
l + 2 ) ] - j 4 .pi. x G [ w 0 Z x ( l + 3 ) , w 12 Z x + 12 ( l +
3 ) , w 24 Z x + 24 ( l + 3 ) , , w T Z x + T ( l + 3 ) ] - j6.pi.
x G } , H x , 2 = { [ w 6 Z x + 6 ( l ) , w 18 Z x + 18 ( l ) , w
30 Z x + 30 ( l ) , , w T + 6 Z x + T + 6 ( l ) ] [ w 9 Z x + 9 ( l
+ 1 ) , w 21 Z x + 21 ( l + 1 ) , w 33 Z x + 33 ( l + 1 ) , , w T +
9 Z x + T + 9 ( l + 1 ) ] - j2.pi. x G [ w 0 Z x ( l + 2 ) , w 12 Z
x + 12 ( l + 2 ) , w 24 Z x + 24 ( l + 2 ) , , w T Z x + T ( l + 2
) ] - j 4 .pi. x G [ w 3 Z x + 3 ( l + 3 ) , w 15 Z x + 15 ( l + 3
) , w 27 Z x + 27 ( l + 3 ) , , w T + 3 Z x + T + 3 ( l + 3 ) ] -
j6.pi. x G } , H x , 3 = { [ w 9 Z x + 9 ( l ) , w 21 Z x + 21 ( l
) , w 33 Z x + 33 ( l ) , , w T + 9 Z x + T + 9 ( l ) ] [ w 0 Z x (
l + 1 ) , w 12 Z x + 12 ( l + 1 ) , w 24 Z x + 24 ( l + 1 ) , , w T
Z x + T ( l + 1 ) ] - j2.pi. x G [ w 3 Z x + 3 ( l + 2 ) , w 15 Z x
+ 15 ( l + 2 ) , w 27 Z x + 27 ( l + 2 ) , , w T + 3 Z x + T + 3 (
l + 2 ) ] - j 4 .pi. x G [ w 6 Z x + 6 ( l + 3 ) , w 18 Z x + 18 (
l + 3 ) , w 30 Z x + 30 ( l + 3 ) , , w T + 6 Z x + T + 6 ( l + 3 )
] - j6.pi. x G } , Eq ( 6 ) ##EQU00001##
where T=12(S-1) is the index of the last subband in interlace
0.
[0054] Each hypothesis H.sub.x,y in equation (6) includes four rows
of channel gains, one row for each OFDM symbol. Each row includes S
channel gains for S pilot subbands in one OFDM symbol. The channel
gains are derived from the received symbols that are extracted from
different subbands depending on the frequency bin error x and the
pilot offset y.
[0055] In an embodiment, a metric is derived for frequency/time
hypothesis H.sub.x,y based on a channel impulse response estimate.
For this embodiment, the channel gains from the four OFDM symbols
for hypothesis H.sub.x,y are first sorted based on subband indices.
As an example, for hypothesis H.sub.x,0, the sorted channel gains
may be given as:
{ H x , y } = { H 0 = w 0 Z x ( l ) , H 1 = w 3 Z x + 3 ( l + 1 ) -
j2.pi. x G , H 2 = w 6 Z x + 6 ( l + 2 ) - j4.pi. x G , H 3 = w 9 Z
x + 9 ( l + 3 ) - j6.pi. x G , H 4 = w 12 Z x + 12 ( l ) , H 4 S -
1 = w T + 9 Z x + T + 9 ( l + 3 ) - j6.pi. x G } . Eq ( 7 )
##EQU00002##
[0056] An FFT/DFT may then be performed on the 4S sorted channel
gains {H.sub.x,y} to obtain a time-domain channel impulse response
estimate with 4S channel taps, which may be given as:
{h.sub.x,y}={h.sub.x,y(0),h.sub.x,y(1),h.sub.x,y(2), . . .
,h.sub.x,y(4S-1)}. Eq (8)
Since 4S is not a power of 2, the channel gains {H.sub.x,y} may be
zero-filled to a power two, and an FFT may then be performed on the
zero-filled channel gains.
[0057] In general, channel gains may be obtained for any number of
interlaces and any number of subbands in each interlace. The length
of the channel impulse response estimate is dependent on the number
of channel gains and may be shorter than 4S. As shown in Table 1,
DVB-H has many more scattered pilot subbands than ISDB-T. To reduce
computational complexity, a subset of the scattered pilot subbands
may be used to derive the metric for each hypothesis. For example,
the first 16, 32 and 64 pilot subbands in each OFDM symbol may be
used for modes 1, 2 and 3, respectively, in DVB-H. The channel
impulse response estimate for each hypothesis may then be derived
using 64-, 128- and 256-point FFTs for modes 1, 2 and 3,
respectively. The FFT size is four times the number of pilot
subbands selected for use.
[0058] If hypothesis H.sub.x,y is a wrong hypothesis, then one or
both of the following apply: [0059] 1. The extracted received
symbols are received data symbols having random complex values.
After the PRBS despreading, the despread symbols remain random
complex values. [0060] 2. The extracted received symbols are
received pilot symbols that are shifted from their correct
frequency alignment by a multiple of 3 subbands. When these
received pilot symbols are despread with the PRBS sequence, the
resultant despread symbols are random scrambled values. In either
of the two cases above, the despread symbols are noisy and are not
representative of the channel gains. The channel impulse response
estimate derived from these noisy despread symbols would then
contain mostly noise.
[0061] Conversely, if hypothesis H.sub.x,y is the correct
hypothesis, then the extracted received symbols are the received
pilot symbols properly aligned in both time and frequency. When
these received pilot symbols are despread with the PRBS sequence,
the resultant despread symbols are good estimates of the channel
gains. A channel impulse response estimate may then be derived
based on these channel gains. This channel impulse response
estimate includes a signal component that is above the noise
floor.
[0062] A metric may be defined based on the channel impulse
response estimate in various manners. In an embodiment, a metric
M.sub.x,y.sup.a is set to the energy of the largest tap in the
channel impulse response estimate, which may be expressed as:
M x , y a = max n h x , y ( n ) 2 . Eq ( 9 ) ##EQU00003##
[0063] In another embodiment, a metric M.sub.x,y.sup.b is set to
the total energy of all taps in the channel impulse response
estimate, which may be expressed as:
M x , y b = n = 0 4 S - 1 h x , y ( n ) 2 . Eq ( 10 )
##EQU00004##
[0064] In yet another embodiment, the metric M.sub.x,y.sup.c is set
to the energy of large taps in the channel impulse response
estimate, which may be expressed as:
M x , y c = h x , y ( n ) 2 > E th h x , y ( n ) 2 , Eq ( 11 )
##EQU00005##
where E.sub.th is a threshold used to determine whether a given tap
is large. E.sub.th may be set to a fixed value or to a
predetermined percentage (e.g., 10%) of the total energy of all
taps.
[0065] In yet another embodiment, a metric M.sub.x,y.sup.d is set
to a non-coherent sum of metrics obtained for multiple (L) sets of
OFDM symbols, as follows:
M x , y d = i | M x , y ( i ) , Eq ( 12 ) ##EQU00006##
where M.sub.x,y(i) is a metric obtained for OFDM symbol set i.
M.sub.x,y(i) may be obtained based on equation (9), (10) or (11).
The L OFDM symbol sets may be adjacent to one another or spread out
over time.
[0066] In general, a metric M.sub.x,y may be derived for hypothesis
H.sub.x,y based on equation (9), (10), (11), (12) or some other
equation. For the embodiments described above, the FFT operation
coherently sums the channel gains {H.sub.x,y} and provides the
channel taps {h.sub.x,y}. This coherent sum provides high
processing gain and yields good detection performance even in low
SNR conditions. In some other embodiments, metric M.sub.x,y may be
derived based on the channel gains {H.sub.x,y} in other manners,
e.g., by summing the energies of the channel gains.
[0067] In any case, a metric M.sub.x,y is obtained for each
frequency/time hypothesis. The metrics for all frequency/time
hypotheses may be compared, and the hypothesis with the largest
metric may be provided as the correct hypothesis. The frequency bin
error for the correct hypothesis may be provided to rotator 414, as
shown in FIG. 4. The pilot offset for the correct hypothesis may be
provided to channel estimator 422 and possibly other processing
units within receiver 150.
[0068] FIG. 5 shows a block diagram of a frequency bin error
estimator 420a, which is an embodiment of estimator 420 within OFDM
demodulator 160 in FIG. 4. Within estimator 420a, a control unit
510 receives inputs indicative of the range of frequency errors
(e.g., .+-.4 bins) and whether the pilot offset is known. Control
unit 510 forms hypotheses covering all frequency and/or time
uncertainty. A despreading unit 512 obtains received symbols for
the K total subbands, extracts the received symbols from the proper
subbands for the hypothesis H.sub.x,y being evaluated, performs
despreading of the extracted received symbols with the PBRS
sequence, rotates the despread symbols for each OFDM symbol by
e.sup.-j2.pi.xG to obtain the channel gains {H.sub.x,y}, where v is
0, 1, 2 and 3 for the four OFDM symbols in a set being
evaluated.
[0069] Channel estimator 422 receives the channel gains for each
hypothesis H.sub.x,y and derives a channel impulse response
estimate {h.sub.x,y} for that hypothesis. A metric computation unit
514 derives a metric M.sub.x,y for each hypothesis based on the
channel impulse response estimate, e.g., using any of the
embodiments described above. Unit 514 may non-coherently sum
multiple metrics obtained for different OFDM symbol sets as shown
in equation (12) or may omit this non-coherent sum, e.g., for a
fast fading channel. A detection unit 516 receives the metrics for
all hypotheses, identifies the largest metric, and provides the
hypothesis with the largest metric as the correct hypothesis.
[0070] For frequency bin error estimation scheme 4 in Table 2, the
frequency bin error may be determined based on the continual pilot
that is sent on the same interlace in all OFDM symbol periods so
that there is no ambiguity as to the pilot subbands. Once the
frequency bin error has been determined, the pilot offset may be
ascertained based on the scattered pilot. By decoupling the
frequency bin error and the pilot offset, the frequency bin error
may be determined with 13 frequency hypotheses for a frequency
error range of .+-.6 bins, and the pilot offset may be determined
with 4 time hypotheses.
[0071] A frequency hypothesis H.sub.x corresponds to a hypothesized
frequency error of x bins. The number of frequency hypotheses to
evaluate is dependent on the frequency error range. Each frequency
hypothesis may be evaluated as follows.
[0072] If hypothesis H.sub.x is correct, then continual pilot
symbols are received on subbands k+x, for k.epsilon.CP, where CP
denotes the set of continual pilot subbands to be considered. CP
may contain all or a subset of the continual pilot subbands.
Equations (4) and (5) may then be expressed as:
Z k + x ( l ) = H k ( l ) 4 3 w k + N k + x ( l ) , and Eq ( 13 ) Z
k + x ( l + 1 ) = j 2 .pi. x G H k ( l + 1 ) 4 3 w k + N k + x ( l
+ 1 ) , where S k ( l + 1 ) = S k ( l ) = 4 3 w k Eq ( 14 )
##EQU00007##
are pilot symbols sent on subband k. Since the same PBRS sequence
is used for all OFDM symbols, the pilot symbols are not a function
of OFDM symbol index l.
[0073] If the wireless channel is relatively static over two
consecutive OFDM symbol periods, then
H.sub.k(l+1).apprxeq.H.sub.k(l) for all subbands. In this case, the
correlation between two received symbols in two OFDM symbols l and
l+1 for each pilot subband may be expressed as:
Z k + x ( l ) Z k + x * ( l + 1 ) .apprxeq. 16 9 - j2.pi. x G H k 2
. Eq ( 15 ) ##EQU00008##
The correlation results may be accumulated across all pilot
subbands, as follows:
k .di-elect cons. CP Z k + x ( l ) Z k + x * ( l + 1 ) .apprxeq. 16
9 - j2.pi. x G k .di-elect cons. CP H k 2 . Eq ( 16 )
##EQU00009##
The correlation results may further be accumulated across multiple
correlation intervals, as follows:
l k .di-elect cons. CP Z k + x ( l ) Z k + x * ( l + 1 ) .apprxeq.
16 9 - j2.pi. x G l k .di-elect cons. CP H k 2 . Eq ( 17 )
##EQU00010##
Each correlation interval corresponds to a different pair of OFDM
symbols. For example, a first accumulated result may be obtained
for OFDM symbols l and l+1 as shown in equation (16), a second
accumulated result may be obtained for OFDM symbols l+1 and l+2, a
third accumulated result may be obtained for OFDM symbols l+2 and
l+3, and a fourth accumulated result may be obtained for OFDM
symbols l+3 and l+4. The four accumulated results may then be
summed to obtain the overall result shown in equation (17). In
general, the correlation results may be accumulated across any
number of subbands and any number of OFDM symbols.
[0074] If hypothesis H.sub.x is not correct because the
hypothesized frequency bin error x is not equal to the actual
frequency bin error m, or x.noteq.m, then received data symbols are
extracted from subbands k+x, for k.epsilon.CP. Equations (4) and
(5) may then be expressed as:
Z.sub.k+x(l)=H.sub.k+x-m(l)D.sub.k+x-m(l)+N.sub.k+x-m(l), and Eq
(18)
Z.sub.k+x(l+1)=e.sup.-j2.pi.xGH.sub.k+x-m(l+1)D.sub.k+x-m(l+1)+N.sub.k+x-
(l+1), Eq (19)
where D.sub.k+x-m(l) and D.sub.k+x-m(l+1) are data symbols sent on
subband k+x-m in OFDM symbols l and l+1, respectively. The
extracted received symbols may be correlated and accumulated across
pilot subbands, as follows:
k .di-elect cons. CP Z k + x ( l ) Z k + x * ( l + 1 ) .apprxeq. 16
9 - j2.pi. m G k .di-elect cons. CP H k + x - m 2 D k + x - m ( l )
D k + x - m * ( l + 1 ) . Eq ( 20 ) ##EQU00011##
Equation (20) indicates that the magnitude squares of the channel
gains do not sum up coherently due to the random nature of the data
symbols D.sub.k+x-m(l) and D.sub.k+x-m(l+1). If the data symbols
are independently and identically distributed (i.i.d.) with zero
mean, which is typically the case, then the accumulated result may
be given as:
l k .di-elect cons. CP Z k + x ( l ) Z k + x * ( l + 1 ) .fwdarw.
0. Eq ( 21 ) ##EQU00012##
Equation (21) indicates that the accumulated result approaches zero
if the accumulation is performed over a sufficient number of OFDM
symbols.
[0075] A metric Q.sub.x.sup.a may be defined for hypothesis H.sub.x
as follows:
Q x a = Re { j2.pi. x G l k .di-elect cons. CP Z k + x ( l ) Z k +
x * ( l + 1 ) } . Eq ( 22 ) ##EQU00013##
In equation (22), the correlation results
Z.sub.k+x(l)Z.sub.k+x*(l+1) are coherently summed over both
frequency and time, the accumulated result is rotated by
e.sup.-j2.pi.xG, and the real part of the rotated result is
provided as metric Q.sub.x.sup.a. If hypothesis H.sub.x is correct,
then the rotated result would have a large positive real part, and
metric Q.sub.x.sup.a is a large value. Conversely, if hypothesis
H.sub.x is incorrect, then the rotated result is a small value, and
metric Q.sub.x.sup.a is likewise a small value.
[0076] The description above assumes that the wireless channel is
relatively static over the correlation interval. This assumption
may not be true for a fast fading channel, and the correlation
between the received symbols may then be expressed as:
Z k + x ( l ) Z k + x * ( l + 1 ) .apprxeq. 16 9 - j2.pi. x G H k (
l ) H k ( l + 1 ) - j.theta. k ( l ) , Eq ( 23 ) ##EQU00014##
where .theta..sub.k(l) is a random variable for the phase
difference in the wireless channel observed by subband k between
OFDM symbol periods l and l+1. A computer simulation was performed
for different channel realizations and for a number of OFDM symbol
periods. For each OFDM symbol period, the phase difference was
determined for each pilot subband, and the phase differences for
all pilot subbands were plotted as a histogram. This histogram
typically has a single nodal peak.
[0077] If .theta..sub.k(l) is centered near 90.degree., 180.degree.
or 270.degree., then the following metric Q.sub.x.sup.s provides
good performance:
Q x s = l k .di-elect cons. CP Z k + x ( l ) Z k + x * ( l + 1 ) 2
. Eq ( 24 ) ##EQU00015##
In equation (24), the correlation results are coherently summed
over both frequency and time, and the squared magnitude of the
accumulated result is provided as metric Q.sub.x.sup.s.
[0078] In a fast fading channel, the single nodal peak may shift
rapidly from one correlation interval to the next. For example, the
peak may be centered near 0.degree. in one correlation interval and
may shift to 180.degree. in the next correlation interval.
.theta..sub.k(l) may thus be nearly out of phase in consecutive
correlation intervals. In this case, a metric Q.sub.x.sup.f may be
defined as:
Q x f = l k .di-elect cons. CP Z k + x ( l ) Z k + x * ( l + 1 ) 2
. Eq ( 25 ) ##EQU00016##
In equation (25), the correlation results are (1) coherently summed
over frequency to take advantage of the single nodal distribution
of .theta..sub.k(l) for the pilot subbands and (2) non-coherently
summed over time to account for fast and random changes in
.theta..sub.k(l). Metric Q.sub.x.sup.f may provide better
performance for a fast fading channel.
[0079] In general, metric Q.sub.x.sup.s is better for static and
slow fading channels, and metric Q.sub.x.sup.f is better for a fast
fading channel. A metric Q.sub.x.sup.c may be defined based on both
Q.sub.x.sup.s and Q.sub.x.sup.f, as follows:
Q.sub.x.sup.c=.alpha.Q.sub.x.sup.s+(1-.alpha.)Q.sub.x.sup.f, Eq
(26)
where .alpha. is a weighting factor that determines the weights to
be given to Q.sub.x.sup.s and Q.sub.x.sup.f. Q.sub.x.sup.c is equal
to Q.sub.x.sup.s for .alpha.=1, is equal to Q.sub.x.sup.f for
.alpha.=0, and is equal to a weighted sum of Q.sub.x.sup.s and
Q.sub.x.sup.f for 0<.alpha.<1. Computer simulation shows that
.alpha.=0.2 provides good performance for both slow and fast fading
channels. .alpha. may also be a configurable value.
[0080] In general, a metric Q may be derived for hypothesis H.sub.x
based on equation (22), (24), (25), (26) or some other equation.
Metric Q.sub.x may be computed for each frequency hypothesis, and
the metrics for all hypotheses may be compared. The hypothesis with
the largest metric may be provided as the correct hypothesis, as
follows:
m ^ = arg x max { Q x } . Eq ( 27 ) ##EQU00017##
[0081] The pilot offset may be determined based on the scattered
pilot once the frequency bin error has been determined based on the
continual pilot. A time hypothesis H.sub.y corresponds to a
hypothesized pilot offset of y, which means that the scattered
pilot is being sent on interlace 3y in OFDM symbol period l. Four
time hypotheses are formed for y=0, 1, 2 and 3, and each hypothesis
may be evaluated as follows. For hypothesis H.sub.y, the scattered
pilot is hypothesized to have been sent on subbands {circumflex
over (m)}+3y+12j, for j=0, 1, 2 . . . , in OFDM symbol periods l
and l+4. The correlation between two received symbols in OFDM
symbol periods l and l+4 for each pilot symbol may then be
expressed as:
Z.sub.{circumflex over (m)}+3y+12j(l)Z.sub.{circumflex over
(m)}+3y+12j*(l+4). Eq (28)
A metric Q.sub.y may be derived for hypothesis H.sub.y by
substituting Z.sub.{circumflex over (m)}+3y+12j(l)Z.sub.{circumflex
over (m)}+3y+12j*(l+4) for Z.sub.k+x(l)Z.sub.k+x*(l+1) in the
equations described above. Four metrics are obtained for four time
hypotheses. The time hypothesis with the largest metric may be
provided as the correct hypothesis.
[0082] FIG. 6 shows a block diagram of a frequency bin error
estimator 420b, which is another embodiment of estimator 420 within
OFDM demodulator 160 in FIG. 4. Within estimator 420b, a control
unit 610 receives inputs indicative of the range of frequency
errors (e.g., .+-.4 bins) and whether the pilot offset is known.
Control unit 610 forms a set of frequency hypotheses covering all
frequency uncertainty and a set of time hypotheses covering all
time uncertainty. A correlation unit 612 obtains received symbols
for the K total subbands, extracts the received symbols from the
proper subbands for hypothesis H.sub.x or H.sub.y being evaluated,
performs correlation on the extracted received symbols, and
provides correlation results for different subbands and correlation
intervals.
[0083] A metric computation unit 614 derives a metric Q.sub.x or
Q.sub.y for each hypothesis based on the correlation results for
that hypothesis, e.g., using any of the embodiments described
above. Unit 614 may coherently sum the correlation results across
subbands and may coherently or non-coherently sum across
correlation intervals. A detection unit 616 receives the metrics
for all frequency hypotheses, identifies the largest metric, and
provides the frequency bin error for the frequency hypothesis with
the largest metric as the estimated frequency bin error. Detection
unit 616 also receives the metrics for all time hypotheses,
identifies the largest metric, and provides the pilot offset for
the time hypothesis with the largest metric as the correct pilot
offset.
[0084] For scheme 2 in Table 2, hypotheses are formed jointly for
frequency bin error and pilot offset, and each hypothesis is
evaluated using correlation-based metrics, e.g., the metrics shown
in equations (22), (24), (25) and/or (26). For scheme 3 in Table 2,
hypotheses are formed separately for frequency bin error and pilot
offset, and each hypothesis is evaluated using despreading-based
metrics, e.g., the metrics shown in equations (9), (10), (11)
and/or (12). A scheme may also use a combination of
despreading-based metric and correlation-based metric. For example,
a despreading-based metric may be used for frequency hypotheses,
and a correlation-based metric may be used for time hypotheses.
Other metrics defined in other manners may also be used to evaluate
the hypotheses.
[0085] FIG. 7 shows an embodiment of a process 700 for performing
frequency error estimation by despreading the received symbols.
Time-domain input samples are processed to obtain frequency-domain
received symbols for the K total subbands (block 710). Multiple
hypotheses are formed for different frequency bin errors (or bin
offsets), different pilot offsets, or different combinations of
frequency bin error and pilot offset (block 712). For each
hypothesis, received symbols are extracted from the proper subbands
determined by the hypothesis (block 714). The extracted received
symbols are hypothesized to be for (1) a scattered pilot that is
sent on different sets of subbands in different symbol periods
and/or (2) a continual pilot that is sent on the same set of
subbands in all symbol periods. The extracted received symbols for
each hypothesis are despread with a scrambling sequence, e.g., the
PBRS sequence, to obtain despread symbols for that hypothesis
(block 716). A metric is then derived for each hypothesis based on
the despread symbols for that hypothesis (block 718). For block
718, a channel impulse response estimate may be derived for each
hypothesis based on the despread symbols for the hypothesis. The
metric for each hypothesis may then be derived based on the channel
impulse response estimate for the hypothesis, as described above.
In any case, the frequency bin error and/or the pilot offset are
determined based on the metrics for all hypotheses evaluated (block
720).
[0086] FIG. 8 shows an embodiment of a process 800 for performing
frequency error estimation by correlating the received symbols.
Time-domain input samples are processed to obtain frequency-domain
received symbols for the K total subbands (block 810). Multiple
hypotheses are formed for different frequency bin errors, different
pilot offsets, or different combinations of frequency bin error and
pilot offset (block 812). For each hypothesis, received symbols in
multiple symbol periods are extracted from the proper subbands
determined by the hypothesis (block 814). The extracted received
symbols are hypothesized to be for a scattered pilot and/or a
continual pilot. For each hypothesis, correlation is performed on
the extracted received symbols for each subband to obtain
correlation results for that hypothesis (block 816). A metric is
then derived for each hypothesis based on the correlation results
for all subbands and correlation intervals for that hypothesis
(block 818). For example, the metric for each hypothesis may be
derived by coherently summing the correlation results across
subbands and coherently or non-coherently summing the correlation
results across correlation intervals. The metric may also be
derived based on a weighted sum of metrics obtained with different
accumulation schemes, e.g., as shown in equation (26). In any case,
the frequency bin error and/or the pilot phase is determined based
on the metrics for all hypotheses evaluated (block 820).
[0087] FIG. 9 shows an embodiment of a process 900 for performing
frequency error estimation in multiple stages. A frequency error is
determined based on a first pilot (e.g., a continual pilot) by
evaluating a first set of hypotheses for a range of frequency
errors (block 912). A pilot offset is determined based on a second
pilot (e.g., a scattered pilot) by evaluating a second set of
hypotheses for a set of pilot offsets and with the frequency error
determined from the first pilot (block 914). The two sets of
hypotheses may be evaluated using the same or different
metrics.
[0088] Process 700, 800, and/or 900 may be performed by frequency
bin error estimator 420 in FIG. 4, controller/processor 180 in FIG.
1, and/or some other processing unit at receiver 150.
[0089] The techniques described herein may be implemented by
various means. For example, these techniques may be implemented in
hardware, firmware, software, or a combination thereof. For a
hardware implementation, the processing units used to perform
frequency error estimation may be implemented within one or more
application specific integrated circuits (ASICs), digital signal
processors (DSPs), digital signal processing devices (DSPDs),
programmable logic devices (PLDs), field programmable gate arrays
(FPGAs), processors, controllers, micro-controllers,
microprocessors, electronic devices, other electronic units
designed to perform the functions described herein, or a
combination thereof.
[0090] For a firmware and/or software implementation, the
techniques may be implemented with modules (e.g., procedures,
functions, and so on) that perform the functions described herein.
The software codes may be stored in a memory (e.g., memory 182 in
FIG. 1) and executed by a processor (e.g., processor 180). The
memory may be implemented within the processor or external to the
processor.
[0091] The previous description of the disclosed embodiments is
provided to enable any person skilled in the art to make or use the
present invention. Various modifications to these embodiments will
be readily apparent to those skilled in the art, and the generic
principles defined herein may be applied to other embodiments
without departing from the spirit or scope of the invention. Thus,
the present invention is not intended to be limited to the
embodiments shown herein but is to be accorded the widest scope
consistent with the principles and novel features disclosed
herein.
* * * * *