Global variance parameter based on mutual information in a data channel

A method comprising:

• receiving, by a soft output detector, an input signal;
• modifying, using a global variance parameter, a plurality of variance terms in a soft output detector to scale the plurality of variance terms;
• processing, by the soft output detector and using the modified plurality of variance terms, the input signal through a trellis of state probabilities to determine data bits and corresponding soft information;
• and
• outputting, by the soft output detector and to a parity-based decoder circuit, the data bits and corresponding soft information for decoding encoded data from the input signal.

A system comprising:

• means, comprising a plurality of branch variance terms for determining data bits from an input signal, for:
• receiving the input signal;
• modifying a plurality of variance terms using a global variance parameter to scale the plurality of variance terms, wherein the global variance parameter is based on mutual information values;
• processing, using the modified plurality of variance terms, the input signal through a trellis of state probabilities to determine the data bits and corresponding soft information;
• and
• outputting, to a parity-based decoder circuit, the data bits and corresponding soft information for decoding encoded data from the input signal;
• and
• means for:
• receiving a known data pattern for at least one data symbol in the input signal;
• receiving a corresponding cost vector from a cost generator circuit for the known data pattern;
• calculating a symbol mutual information value based on the known data pattern and corresponding cost vector;
• and
• averaging symbol mutual information values for a plurality of data symbols in the input signal across a codeword to determine the mutual information values.