Soft decision maximum likelihood encoder and decoder

A full information set soft decision decoder for correcting b+1 errors using a block erasure correcting code with the ability to correct b erasures, comprising:

• a receiving mechanism adapted to receive for each symbol in a block of n symbols, a set of likelihoods wherein each possible symbol value has a likelihood associated therewith, and a hard decision equal to the symbol value corresponding to a maximum likelihood;
• a converter for converting likelihoods values to log probability ratios;
• an erasure generator adapted to generate all possible combinations of erasure patterns;
• an erasure correction decoder adapted to correct said block in accordance with each erasure pattern generated;
• a likelihood generator for calculating a likelihood of a resulting codeword corresponding to each possible combination of erasure pattern utilizing said log probability ratios;
• search means for determining a codeword yielding a maximum likelihood;
• and
• a decoder mechanism operative to decode said codeword corresponding to a maximum likelihood.

A method of soft decoding capable of correcting b+1 errors using a block erasure correcting code with the ability to correct b erasures, said method comprising the steps of:

• receiving a hard decision metric and a symbol metric for all possible values of each symbol in a codeword;
• assuming a single error in each symbol location;
• and for each symbol,
• calculating a first metric difference between a symbol metric for a first trial symbol and said hard decision metric;
• assuming b erasures in said codeword;
• and for each erasure,
• correcting said erasures using said erasure correcting code;
• calculating a second metric difference between a symbol metric for a second trial symbol and said hard decision metric;
• summing said second metric differences and said first metric difference to yield a sum of differences;
• selecting a codeword yielding a minimum sum of differences;
• and
• wherein b is as positive integer.

A method of generating a soft decision error correcting code, said method comprising the steps of:

• initializing a parity symbol;
• receiving a block of data comprising n−1 symbols;
• and for each symbol, XORing said symbol with said parity symbol;
• rotating said parity symbol;
• including a resultant parity symbol as an nth symbol in said block;
• modulating said block of data and said soft decision error correcting code utilizing Code Shift Keying (CSK) modulation;
• and
• wherein n is a positive integer.

A method of generating a soft decision error correcting code, said method comprising the steps of:

• initializing a horizontal parity symbol;
• initializing a diagonal parity symbol;
• receiving a block of data comprising a plurality of symbols;
• and for each symbol, XORing said symbol with said horizontal parity symbol;
• XORing said symbol with said diagonal parity symbol;
• rotating said diagonal parity symbol;
• calculating an ‘S’ bit from said data block;
• setting said diagonal parity symbol in accordance with said ‘S’ bit;
• appending the resultant horizontal parity symbol to said block;
• modulating said block of data and said soft decision error correcting code utilizing Code Shift Keying (CSK) modulation;
• and
• appending the resultant diagonal parity symbol to said block.

A communication apparatus coupled to a channel, comprising:

• an encoder adapted to generate a soft decision code from a block of data received from a data source;
• a transmitter operative to transmit data including said soft decision code over said channel;
• a receiver operative to receive data from said communications channel and to generate a set of soft symbol metrics and a hard symbol metric for each symbol received;
• a soft decoder adapted to decode said received data so as to yield the original transmitted symbols, said soft decoder operative to:
• receive said hard decision metric and said set of symbol metrics;
• assume a single error in each symbol location;
• and for each symbol,
• calculate a first metric difference between the metric for a first trial symbol and said hard decision metric;
• assume one or more erasures in said symbol;
• and for each erased symbol,
• correct said erasures using an erasure correcting code;
• calculate a second metric difference between the metric for a second trial symbol and said hard decision metric;
• sum said second metric differences and said first metric difference to yield a sum of differences;
• and
• select the codeword yielding a minimum sum of differences.