Pulse or digital communications – Receivers – Particular pulse demodulator or detector
Reexamination Certificate
2000-02-08
2003-09-23
Tran, Khai (Department: 2631)
Pulse or digital communications
Receivers
Particular pulse demodulator or detector
C714S794000
Reexamination Certificate
active
06625236
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates to digital communication methods and systems and more particularly to methods and systems for decoding symbols from a signal that is received on a communications channel.
Digital communication systems and methods including digital wireless communication systems and methods are widely used to convey information. As is well known to those having skill in the art, in digital communications, information is often generated as, or translated into, a stream of bits. A transmitter then maps this bit stream into a symbol stream which is modulated and transmitted on a communications channel. A digital receiver detects the signal and maps the signal back into decoded bits.
As is also well known, equalizers are used to demodulate information symbols from information-modulated signals that are received from a transmitter through a communications channel suffering from time dispersion, giving rise to Inter-Symbol Interference (ISI). Conventional equalizers include transversal equalizers, which apply transversal or finite impulse response filters to the received signal which are the inverse of the channel characteristic, thus undoing at least some of the deleterious effect of the channel. Unfortunately, transversal equalizers may not be effective for all types of channels, and can result in magnification of noise.
Another known type of equalizer is a Decision Feedback Equalizer (DFE) wherein already decoded bits are used to subtract out their delayed echoes caused by time dispersion, thereby canceling ISI and enabling the next bit to be decoded. Unfortunately, DFE equalizers may only work well when later delayed echoes are weaker than the signal component being decoded. Otherwise, DFE may discard too much useful energy in the echoes. It is possible in such circumstances to store received signal samples and process them retrospectively in time reversed order. Equalizers that process in either a forward or reverse time order are described, for example, in U.S. Pat. Nos. 5,841,816 to coinventor Dent and Croft entitled “Diversity PI/4-DQPSK Demodulation” and U.S. Pat. No. 5,335,250 to coinventor Dent and Chennakeshu, entitled “Method and Apparatus for Bidirectional Demodulation of Digitally Modulated Signals”, the disclosures of which are hereby incorporated herein by reference in their entirety. These patents also describe a Maximum Likelihood Sequence Estimation (MLSE) technique which, unlike DFE, need not discard energy in delayed echoes.
Equalizers generally may need to determine the phase and magnitude of all significant delayed echoes relative to some nominal delayed ray, which values characterize the transmission channel and therefore are called channel coefficients or channel estimates. Channel estimates may be determined by including known symbols in the transmitted signal in the form of a “syncword” or “pilot symbols”. Alternatively, in the case of equalizers known as blind equalizers, the channel and the unknown data may be estimated simultaneously. One form of blind equalizer is described in U.S. Pat. No. 5,557,645 to coinventor Dent, entitled “Channel-Independent Equalizer Device” the disclosure of which is hereby incorporated herein by reference in its entirety.
When an equalizer or demodulator is to be followed by an error correction decoder, it can be advantageous for error correction decoding to pass “soft” decisions rather than “hard” decisions from the demodulator to the error correction decoder. The soft decisions should preferably be values indicative of the likelihood that a particular symbol is one or other value, and more specifically, preferably proportional to the negative logarithm of the symbol likelihood. Often, the signal-to-noise ratio of the symbol is an adequate soft decision. Such soft decisions may be obtained from MLSE equalizers by a technique described in U.S. Pat. No. 5,099,499 to Hammar, entitled “Method of Generating Quality Factors for Binary Digits Obtained in the Viterbi-Analysis of a Signal” the disclosure of which is hereby incorporated herein by reference in its entirety. Hammar equates a soft decision for a binary symbol to a difference between a metric for the opposite value of the symbol minus the metric for the decoded value of the symbol, divided by the sum of all metrics, or final metric, of the MLSE equalizer. The difference metrics are indicative of the symbol's signal strength while the final metric is indicative of the noise in the channel. Thus Hammar's soft information can be a measure of signal-to-noise ratio on a per-symbol basis.
U.S. Pat. No. 5,331,666 to coinventor Dent, entitled “Adaptive Maximum Likelihood Demodulator” the disclosure of which is hereby incorporated herein by reference in its entirety, describes an MLSE device for four-phase modulations such as QPSK, OQPSK, DQPSK and &pgr;/4-DQPSK. It describes how metric computations may be simplified by exploiting quadrantal symmetry in a 4-phase constellation.
In co-pending application Ser. No. 09/499,977, filed Feb. 8, 2000, entitled Methods, Receivers And Equalizers For 8PSK Modulation Having Increased Computational Efficiency to coinventor Zangi, an 8-PSK equalizer is described that exploits octant-symmetries in the constellation to reduce metric calculations. This application is hereby incorporated by reference herein in its entirety.
Unfortunately, the above-described equalization techniques may become computationally intensive when the number of symbols and/or the amount of intersymbol interference becomes large. The computational complexity may result in increased power consumption in a mobile terminal and/or increased processing time. Moreover, the symbol decisions that are made in an equalizer may not provide a desired input for an error correction decoder. Thus, in order to increase the probability that individual symbols are decoded with high probability, further complexity may be introduced.
In particular, for various reasons, symbol decisions made in an equalizer may not provide a desired input for an error correction decoder. An MLSE equalizer determines a sequence which, with highest probability, is the sequence that was transmitted. However, an individual symbol may not be the symbol which, with highest probability, was the symbol transmitted. To obtain the probability of individual symbols, a different type of demodulation known as Maximum A-Posteriori (MAP) decoding may be used, which however can be of excessive complexity. A difference between MAP and MLSE derives from the tendency of error events in an MLSE device to be multiple error events, in which a group of adjacent symbols may be erroneous or of high error probability.
Moreover, full MLSE often is too complex and a combination of MLSE and DFE may be used, known as Per-Survivor Processing (PSP). MLSE-PSP may be useful when the time dispersion spans a large number L of symbol periods, the symbols are from a larger than binary alphabet of M symbols, and the resulting full MLSE complexity, which is proportional to M
L
may be too large. For example, for 8-PSK (M=8) and five symbol periods of time dispersion, an MLSE device may need to compute 8
5
=32,768 metrics per decoded symbol, which generally is excessive.
In U.S. application Ser. No. 09/237,356 filed Jan. 26, 1999 entitled “Reduced Complexity MLSE Equalizer for M-ARY Modulated Signals” to coinventor Zangi, the disclosure of which is hereby incorporated herein by reference in its entirety, it is shown that this number may be reduced by exploiting octant symmetries in 8-PSK. However, even reducing metric computations four or eightfold may not be enough. As a result, less than full MLSE may need to be used.
In reduced MLSE, not all combinations of five successive symbols (for 5-symbol time dispersion) are computed. The oldest (for example three) symbols are instead “decided” and their values used in a DFE operation to subtract out the ISI caused by the already decided symbols, leaving only, for example, two symbols to be hypothesized in the MLSE part of the device. This can reduc
Dent Paul W.
Zangi Kambiz
Ericsson Inc.
Myers Bigel & Sibley & Sajovec
Tran Khai
LandOfFree
Methods and systems for decoding symbols by combining... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and systems for decoding symbols by combining..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and systems for decoding symbols by combining... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3005963