Metric corrections for demodulators using serial localization with indecision
AssigneeTelefonaktiebolaget L M Ericsson (publ)
InventorsAli S. Khayrallah, Dennis Hui
Life of this patent
Estimating the term…
Abstract
Demodulation methods and apparatus for a multi-stage SLI demodulator are disclosed. Residual signals from each demodulation stage are modeled as finite sets of unresolved signals and a new metric is introduced for use in search of best candidate symbol estimates. The metric may be evaluated based on a probability distribution function of the residual signals or a probability mass function of the unresolved signals. The metric may also be approximated by the sum of a conventional Euclidean metric and a correction metric. The best candidate symbol estimates generated from each stage of the multi-stage SLI demodulator are summed to form cumulative symbol estimates.
Claims20
A method of demodulating multiple data streams implemented in one demodulation stage of a multi-stage receiver, said method comprising:
- receiving input symbols corresponding to multiple data streams;
- generating a set of candidate sequences from symbols in a symbol constellation, and a finite set of unresolved signals obtained by modulating the set of candidate sequences and subtracting the modulated candidate sequences from the input symbols;
- computing a likelihood metric for each candidate sequence based on the input symbols and a probability mass function for the unresolved signals;
- and
- selecting, based on the likelihood metrics, one of said candidate sequences as output symbol estimates to be summed with output symbol estimates from one or more other demodulation stages to generate final estimates of the multiple data streams.
A demodulator for a multi-stage receiver having a plurality of demodulation stages, said demodulator comprising:
- an input to receive input symbols corresponding to multiple data streams and a symbol constellation;
- a candidate symbol generator to generate a set of candidate sequences from symbols in said symbol constellation, and to generate a finite set of unresolved signals obtained by modulating the set of candidate sequences and subtracting the modulated candidate sequences from the input symbols;
- a likelihood estimator to compute a likelihood metric for each candidate sequence based on the input symbols and a probability mass function for the unresolved signals;
- and
- a selector to select one of said candidate sequences based on the likelihood metrics computed for the set of candidate sequences as output symbol estimates to be summed with output symbol estimates from one or more other demodulation stages to generate final estimates of the multiple data streams.
Dependencies are inferred from the claim text.