Diversity receivers and methods for relatively-delayed signals
View Patent ↗Methods, software and apparatuses for decoding received delayed path signals are described. A receiver receives a signal via different paths with relative delay, e.g., on the same frequency channel. The signal is sampled at at least one sample per symbol. A channel estimator determines channel coefficients associated with the channel over which the signals were transmitted. A processor selects samples that include a dependence on a particular symbol being decoded and computes an error metric for each possible value of the particular symbol, for combinations of hypothesized future symbol values on which the selected samples depend, and for single values of previous symbols on which the selected samples depend. Error metrics calculated for each symbol are stored in memory. A soft symbol is generated based on said signal samples from the sampler, the channel coefficients and estimates of symbols computed.
1. A receiver for processing a signal modulated with data symbols and received via a plurality of paths having a relative delay therebetween, the receiver comprising:
a sampler configured to sample said signal to obtain at least one sample per data symbol time interval;
a channel estimator configured to determine a set of complex channel coefficients that characterize said delayed paths;
a metric computer configured to compute an estimate of a current symbol and an associated metric, wherein said current symbol estimate and said associated metric are based on signal samples from said sampler, said channel coefficients from said channel estimator, estimates of previously estimated symbols and associated metrics, and hypotheses of not yet estimated symbols, and wherein the metric computer is further configured to store said current symbol estimates and said associated metrics associated with combinations of said hypothesized symbols in a metric table for said current symbol; and
a soft symbol value generator configured to generate a soft symbol value for a symbol based on said signal samples from said sampler, said channel coefficients from said channel estimator and estimates of symbols computed by said metric computer.
2. The receiver of claim 1 , wherein at least one of said estimates of previously estimated symbols corresponds to a previously estimated symbol stored in its associated metric table having a lowest metric of all metrics with the same value of hypothesized symbols as symbols currently being hypothesized by said metric computer.
3. The receiver of claim 1 , wherein said current symbol, at least one previously estimated symbol and at least one hypothesized symbol are used to specify hypothesized values of those symbols in a previously computed metric table in order to obtain estimates of other previously estimated symbols needed to compute said metric table for the current symbol.
4. The receiver of claim 1 , further comprising a decision refinement processor, which substitutes symbol estimates made subsequently by said metric computer for previously hypothesized symbols in order to select a reduced set of entries in a previously computed metric table.
5. The receiver of claim 4 , wherein when selecting said reduced number of entries in said previously computed metric table, if said decision refinement processor determines a non-selected table entry was used as an estimate of an already estimated symbol by said metric computer in computing an entry in a later metric table said metric computer re-computes said later metric table entry if said already estimated symbol would now have a different value than the value used previously.
6. The receiver of claim 1 , further comprising an error correction decoder for decoding said soft symbol values to produce error corrected symbols.
7. The receiver of claim 6 , further comprising an error detection decoder to determine if said error correction decoder was successful in correcting symbol errors.
8. The receiver of claim 7 , wherein when said error detection decoder determines that said error correction decoder did not successfully correct all symbol errors, the best estimates of error corrected symbols from said error correction decoder are substituted for symbols estimated by said metric computer and used by said soft symbol generator to generate new soft symbol values for iteratively repeating said error correction decoding and error detection decoding.
9. The receiver of claim 1 , wherein said receiver is configured for filtering, amplifying, and analog-to-digital converting said summed signal to obtain said at least one sample per data symbol time interval.
10. The receiver of claim 1 , wherein said metric computer stores an index for each row in said metric table indicating a location of one or more values stored in said row.
11. The receiver of claim 10 , wherein said index is used for traceback in determining the final output of the receiver.
12. The receiver of claim 10 , wherein said associated metric is a sum of the squares of residual errors associated with said current symbol estimate.
13. A method for processing a signal modulated with data symbols and received through a plurality of paths, the method comprising:
sampling said signal to obtain at least one sample per data symbol time interval;
determining a set of complex channel coefficients that characterize said delayed paths;
computing an estimate of a current symbol and an associated metric, wherein said current symbol estimate and said associated metric are based on said sampling of said signal, said determining of said set of complex channel coefficients, estimates of previously estimated symbols and associated metrics, and hypotheses of not yet estimated symbols;
storing said current computed symbol estimates and said associated metrics for combinations of said hypothesized symbols in a metric table for said current symbol; and
generating a soft symbol value for a symbol based on said sampled signal, said determined set of channel coefficients and computed estimates of symbols.
14. The method of claim 13 , wherein at least one of said estimates of previously estimated symbols corresponds to a previously estimated symbol stored in its associated metric table having a lowest metric of all metrics with the same value of hypothesized symbols as symbols currently being hypothesized by said metric computer.
15. The method of claim 13 , further comprising specifying hypothesized values of symbols in a previously computed metric table in order to obtain estimates of other previously estimated symbols needed to compute said metric table for said current symbol, wherein said current symbol, at least one previously estimated symbol, and at least one hypothesized symbol are used for said specifying hypothesized values of symbols.
16. The method of claim 13 , further comprising substituting symbol estimates subsequently computed for previously hypothesized symbols in order to select a reduced set of entries in a previously computed metric table.
17. The method of claim 13 , further comprising:
determining if a non-selected table entry was used as an estimate of an already estimated symbol in computing an entry in a later metric table; and
re-computing said later metric table entry if said already estimated symbol would now have a different value than the value used previously.
18. The method of claim 13 , further comprising decoding said soft symbol values to produce error corrected symbols.
19. The method of claim 18 , further comprising determining if said error correction decoding was successful in correcting symbol errors.
20. The method of claim 19 , further comprising substituting the best estimates of error corrected symbols for symbols estimated by computing an estimate to generate new soft symbol values for iteratively repeating said error correction decoding and error detection decoding when said error detection decoding determines that said error correction decoding did not successfully correct all symbol errors.
21. The method of claim 13 , wherein sampling includes filtering, amplifying, and analog-to-digital converting said summed signal to obtain at least one sample per data symbol time interval.
22. The method of claim 13 , further comprising storing an index for each row in said metric table indicating the location of one or more values stored in said row.
23. The method of claim 22 , further comprising using said index for traceback in determining the final output of the receiver.
24. A non-transitory computer-readable medium having stored thereon instructions for decoding a signal modulated with data symbols and received via delayed paths, said instructions comprising code for performing the steps of:
sampling said signal to obtain at least one sample per data symbol time interval;
determining a set of complex channel coefficients that characterize said delayed paths;
hypothesizing not yet estimated symbols;
computing an estimate of a current symbol and an associated metric, wherein said current symbol estimate and said associated metric are based on said sampling, said determining of said set of complex channel coefficients, said hypothesizing of not yet estimated symbols, estimates of previously estimated symbols and associated metrics;
storing said current computed symbol estimates and said associated metrics with combinations of said hypothesized symbols in a metric table for said current symbol; and
generating a soft symbol value for a symbol based on said sampled signal, said determined set of channel coefficients, and computed estimates of symbols.
25. The non-transitory computer-readable medium of claim 24 , wherein at least one of said
estimates of previously estimated symbols corresponds to a previously estimated symbol stored in its associated metric table having the lowest metric of all metrics with the same value of hypothesized symbols as symbols currently being hypothesized by said metric computer.
26. A method for performing equalization in a diversity receiver comprising:
receiving an information signal along at least two different paths, one path having a time delay L relative to the other path;
wherein said time delay L is a non-integer number of symbol periods, and a value of L is greater than 25 symbol periods; and
generating a soft symbol value for a symbol in said information signal based on samples from said information signal, channel coefficients associated with said information signal and stored estimates associated with other symbols in said information signal,
wherein a computational complexity associated with said step of generating is on the order of M, where M is a size of a symbol alphabet associated with said information signal.
27. The method of claim 26 , wherein a delay profile associated with said time delay L is uneven.
28. The method of claim 27 , wherein said uneven delay profile includes a first time dispersion value of a plurality of microseconds associated with a symbol in one of said at least two different paths and a second time dispersion value of a plurality of microseconds associated with a corresponding symbol in the other of said at least two different paths.