US 6,690,739 B1 · Method for intersymbol interference compensation

Abstract

The intersymbol interference compensation (ISIC) method of the invention comprises an encoding step performed by an encoder and a data estimating step performed a data estimator. The encoder performs at least a trellis encoding of a data sequence using a trellis forward-error-correction (FEC) code to provide an encoded sequence, which is supplied to a channel. The data estimator, which advantageously processes a channel output sequence to arrive at an estimate of the data sequence, comprises at least a template generator for generating templates, a partial equalizer for processing the channel output sequence to provide a partially equalized sequence that is substantially free of precursor ISI, and an ISIC decoder for trellis decoding and ISI compensation. The ISIC decoder computes the branch metric for each current branch based on a single-sided template {circumflex over (q)}(x,y) that accounts for the postcursor ISI in the partially equalized sequence, where x is the vector of symbols associated with the past branch or branches that are connected to the current branch and y is the vector of symbols associated with the current branch. The number of states in the ISIC decoder is determined by the trellis code and is independent of the ISI span of the channel. The invention also provides for techniques for adapting the partial equalizer using the templates, estimating the ISI profile, and updating the templates. In one embodiment of the invention, concatenated coding is employed. In yet another embodiment of the invention, the data estimator employs a decision-directed-ISIC (DD-ISIC) decoder that computes the branch metric for each current branch based on a double-sided template ŝ(x,y,z), where z is the vector of symbols associated with the future branch or branches connected to the current branch and is determined by an interim estimate of the data sequence. A double-sided partial equalizer that removes only some of the precursor and postcursor ISI provides the input to the DD-ISIC decoder, which compensates for the unequalized ISI and performs trellis decoding simultaneously.

Claims28

Dependent claims

A method of intersymbol interference compensation (ISIC) for preparing a plurality of data symbols for use as input to a channel and recovering said plurality of data symbols from a plurality of channel output samples for use by a data sink, said method comprising the steps of

encoding said plurality of data symbols to provide encoded symbols for use as input to said channel, said step of encoding further including the steps of
(a) trellis encoding said plurality of data symbols using a trellis code to provide trellis-encoded symbols;
and
(b) mapping said trellis-encoded symbols to provide said encoded symbols such that said encoded symbols assume values from a predetermined alphabet;
and
performing data recovery on said plurality of channel output samples to provide estimated data symbols for use by said data sink, said step of performing data recovery further including the steps of
(a) partially equalizing said plurality of channel output samples to provide partially equalized samples using a partial equalizer, thereby equalizing substantially all of the precursor intersymbol interference (ISI) and at most some of the postcursor ISI in said plurality of channel output samples;
(b) generating a set of single-sided templates based on a noiseless channel model having a truncated ISI profile, wherein each single-sided template {circumflex over (q)}(•,•) is a function of two vector arguments having elements taken from said predetermined alphabet;
and
(c) ISIC decoding said partially equalized samples to provide said estimated data symbols, said step of ISIC decoding further including the step of computing a branch metric for each current branch of a decoder trellis of said trellis code based on the partially equalized samples associated with said current branch and the single-sided template {circumflex over (q)}(x,y) with vector argument x being the vector of a predetermined number of symbols associated with the past branches connected to said current branch and vector argument y being the vector of symbols associated with said current branch;
and the step of using said branch metric in a trellis decoding algorithm.

8 more elements

A method of intersymbol interference compensation for preparing a plurality of data symbols for use as input to a channel and recovering said plurality of data symbols from a plurality of channel output samples for use by a data sink, said method comprising the steps of

encoding said plurality of data symbols to provide encoded symbols for use as input to said channel, said step of encoding further including the steps of
(a) trellis encoding said plurality of data symbols using a trellis code to provide trellis-encoded symbols;
and
(b) mapping said trellis-encoded symbols to provide said encoded symbols such that said encoded symbols assume values from a predetermined alphabet;
and
performing data recovery on said plurality of channel output samples to provide estimated data symbols for use by said data sink, said step of performing data recovery further including the steps of
(a) partially equalizing said plurality of channel output samples to provide partially equalized samples using a partial equalizer, thereby equalizing substantially all of the precursor ISI and at most some of the postcursor ISI in said plurality of channel output samples;
(b) generating a set of single-sided templates based on a first noiseless channel model having a first truncated ISI profile, wherein each single-sided template {circumflex over (q)}(•,•) is a function of two vector arguments having elements taken from said predetermined alphabet;
(c) ISIC decoding said partially equalized samples to provide interim data decisions, said step of ISIC decoding further including the step of computing a first branch metric for each first current branch of a first decoder trellis of said trellis code based on the partially equalized samples associated with said first current branch and the single-sided template {circumflex over (q)}(x1,y1) with vector argument x1 being the vector of a first predetermined number of symbols associated with the past branches connected to said first current branch and vector argument y1 being the vector of symbols associated with said first current branch;
and the step of using said first branch metric in a first trellis decoding algorithm;
(d) applying double-sided partial equalization to said plurality of channel output samples to provide double-sided partially equalized samples using a double-sided partial equalizer, thereby equalizing at most some of the precursor and postcursor ISI in said plurality of channel output samples;
(e) generating a set of double-sided templates based on a second noiseless channel model having a second truncated ISI profile, wherein each double-sided template ŝ(•,•,•) is a function of three vector arguments having elements taken from said predetermined alphabet;
and
(f) decision-directed-ISIC decoding said double-sided partially equalized samples to provide said estimated data symbols, said step of decision-directed-ISIC decoding further including the step of computing a second branch metric for each second current branch of a second decoder trellis of said trellis code based on the double-sided partially equalized samples associated with said second current branch and the double-sided template ŝ(x2,y2,z2) with vector argument x2 being the vector of a second predetermined number of symbols associated with the past branches connected to said second current branch and vector argument y2 being the vector of symbols associated with said second current branch and vector argument z2 being the vector of a third predetermined number of symbols determined using said interim data decisions to be associated with the future branches connected to said second current branch;
and the step of using said second branch metric in a second trellis decoding algorithm.

12 more elements

encoding said plurality of data symbols to provide encoded symbols for use as input to said channel, said step of encoding further including the steps of
(a) outer-FEC encoding said plurality of data symbols using an outer FEC code to provide outer-FEC-encoded symbols;
(b) interleaving said outer-FEC-encoded symbols to provide interleaved symbols;
(c) trellis encoding said interleaved symbols using a trellis code to provide trellis-encoded symbols;
and
(d) mapping said trellis-encoded symbols to provide said encoded symbols such that said encoded symbols assume values from a predetermined alphabet;
and
performing data recovery on said plurality of channel output samples to provide estimated data symbols for use by said data sink, said step of performing data recovery further including the steps of
(a) partially equalizing said plurality of channel output samples to provide partially equalized samples using a partial equalizer, thereby equalizing substantially all of the precursor ISI and at most some of the postcursor ISI in said plurality of channel output samples;
(b) generating a set of single-sided templates based on a first noiseless channel model having a first truncated ISI profile, wherein each single-sided template {circumflex over (q)}(•,•) is a function of two vector arguments having elements taken from said predetermined alphabet;
(c) ISIC decoding said partially equalized samples to provide ISIC-decoded symbols, said step of ISIC decoding further including the step of computing a first branch metric for each first current branch of a first decoder trellis of said trellis code based on the partially equalized samples associated with said first current branch and the single-sided template {circumflex over (q)}(x1,y1) with vector argument x1 being the vector of a first predetermined number symbols associated with the past branches connected to said first current branch and vector argument y1 being the vector of symbols associated with said first current branch;
and the step of using said first branch metric in a first trellis decoding algorithm;
(d) processing said ISIC-decoded symbols to provide interim data decisions;
(e) applying double-sided partial equalization to said plurality of channel output samples to provide double-sided partially equalized samples using a double-sided partial equalizer, thereby equalizing at most some of the precursor and postcursor ISI in said plurality of channel output samples;
(f) generating a set of double-sided templates based on a second noiseless channel model having a second truncated ISI profile, wherein each double-sided template ŝ(•,•,•) is a function of three vector arguments having elements taken from said predetermined alphabet;
(g) decision-directed-ISIC decoding said double-sided partially equalized samples to provide DD-ISIC decoded symbols, said step of decision-directed-ISIC decoding further including the step of computing a second branch metric for each second current branch of a second decoder trellis of said trellis code based on the double-sided partially equalized samples associated with said second current branch and the double-sided template ŝ(x2,y2,z2) with vector argument x2 being the vector of a second predetermined number of symbols associated with the past branches connected to said second current branch and vector argument y2 being the vector of symbols associated with said second current branch and vector argument z2 being the vector of a third predetermined number of symbols determined using said interim data decisions to be associated with the future branches connected to said second current branch;
and the step of using said second branch metric in a second trellis decoding algorithm;
(h) deinterleaving said DD-ISIC decoded symbols to provide deinterleaved symbols;
and
(i) outer-FEC decoding said deinterleaved symbols based on said outer FEC code to provide said estimated data symbols.

17 more elements

encoding said plurality of data symbols to provide encoded symbols for use as input to said channel, said step of encoding further including the steps of
(a) error-detection encoding said plurality of data symbols using an error-detection code to provide ED-encoded symbols;
(b) outer-FEC encoding said ED-encoded symbols using an outer FEC code to provide outer-FEC-encoded symbols;
(c) interleaving said outer-FEC-encoded symbols to provide interleaved symbols;
(d) trellis encoding said interleaved symbols using a trellis code to provide trellis-encoded symbols;
and
(e) mapping said trellis-encoded symbols to provide said encoded symbols such that said encoded symbols assume values from a predetermined alphabet;
and
performing data recovery on said plurality of channel output samples to provide estimated data symbols for use by said data sink, said step of performing data recovery further including the steps of
(a) partially equalizing said plurality of channel output samples to provide partially equalized samples using a partial equalizer, thereby equalizing substantially all of the precursor ISI and at most some of the postcursor ISI in said plurality of channel output samples;
(b) generating a set of single-sided templates based on a first noiseless channel model having a first truncated ISI profile, wherein each single-sided template {circumflex over (q)}(•,•) is a function of two vector arguments having elements taken from said predetermined alphabet;
(c) ISIC decoding said partially equalized samples to provide ISIC-decoded symbols, said step of ISIC decoding further including the step of computing a first branch metric for each first current branch of a first decoder trellis of said trellis code based on the partially equalized samples associated with said first current branch and the single-sided template {circumflex over (q)}(x1,y1) with vector argument x1 being the vector of a first predetermined number of symbols associated with the past branches connected to said first current branch and vector argument y1 being the vector of symbols associated with said first current branch;
and the step of using said first branch metric in a first trellis decoding algorithm;
(d) processing said ISIC-decoded symbols to provide interim data decisions;
(e) applying double-sided partial equalization to said plurality of channel output samples to provide double-sided partially equalized samples using a double-sided partial equalizer, thereby equalizing at most some of the precursor and postcursor ISI in said plurality of channel output samples;
(f) generating a set of double-sided templates based on a second noiseless channel model having a second truncated ISI profile, wherein each double-sided template ŝ(•,•,•) is a function of three vector arguments having elements taken from said predetermined alphabet;
(g) decision-directed-ISIC decoding said double-sided partially equalized samples to provide DD-ISIC decoded symbols, said step of decision-directed-ISIC decoding further including the step of computing a second branch metric for each second current branch of a second decoder trellis of said trellis code based on the double-sided partially equalized samples associated with said second current branch and the double-sided template ŝ(x2,y2,z2) with vector argument x2 being the vector of a second predetermined number of symbols associated with the past branches connected to said second current branch and vector argument y2 being the vector of symbols associated with said second current branch and vector argument z2 being the vector of a third predetermined number of symbols determined using said interim data decisions to be associated with the future branches connected to said second current branch;
and the step of using said second branch metric in a second trellis decoding algorithm;
(h) deinterleaving said DD-ISIC decoded symbols to provide deinterleaved symbols;
(i) outer-FEC decoding said deinterleaved symbols based on said outer FEC code to provide outer-FEC-decoded symbols;
and
(j) error-detection decoding said outer-FEC-decoded symbols based on said error-detection code to provide said estimated data symbols.

19 more elements

Dependencies are inferred from the claim text.

Method for intersymbol interference compensation

Life of this patent

Method for intersymbol interference compensation

Life of this patent

Abstract

Claims28