Maximum likelihood detector and/or decoder

A maximum likelihood detector for receiving a stream of data values which correspond to ideal values but which may include added noise, and outputting information specifying a sequence of states of maximum likelihood selected from a plurality of possible states corresponding to the stream of data values, said ideal values being determined by the possible states, the maximum likelihood detector comprising:

• a pre-processor programmed to obtain or calculate first weighting values indicating likelihoods that each data value corresponds respectively to each of a plurality of said ideal data values;
• a plurality of processors arranged in a hierarchical structure, wherein
• each processor in a lowest level of the hierarchy is programmed to use a plurality of the first weighting values to calculate second weighting values indicating respective likelihoods that a section of the stream of data values corresponds to each of a plurality of possible state sequences, the section comprising one or more possible initial state and one or more possible final states;
• and, for each said possible initial state and each said possible final state, to select a second weighting value of highest likelihood corresponding to a state sequence from the initial state to the final state, and
• each processor in a level above the lowest level of the hierarchy is programmed to use weighting values calculated at a lower level of the hierarchy to calculate third weighting values indicating respective likelihoods that a sequence of data values comprising a plurality of adjacent sections corresponds to each of a plurality of possible state sequences, said sequence of adjacent sections comprising one or more possible initial states and one or more possible final states;
• and, for each said possible initial state and said each possible final state, to select a third weighting value of highest likelihood corresponding to a state sequence from the initial state to the final state;
• and
• an output generator to output information specifying a state sequence of highest likelihood according to the weighting value selections made by the plurality of processors.

A method of maximum likelihood detection for receiving a stream of data values which correspond to ideal values but which may include added noise, and outputting information specifying a sequence of states of maximum likelihood corresponding to the stream of data values, said ideal values being determined by the possible states, the method comprising:

• obtaining first weighting values indicating likelihoods that each data value corresponds respectively to each of a plurality of said ideal data values;
• processing said first weighting values in a hierarchical manner, by:
• in a lowest level of the hierarchy, using a plurality of the first weighting values to calculate second weighting values indicating respective likelihoods that each of a plurality of sections of the stream of data values corresponds to each of a plurality of possible state sequences, each section comprising one or more possible initial state and one or more possible final states;
• and, for each said possible initial state and each said possible final state for each section, selecting a second weighting value of highest likelihood corresponding to a state sequence from the initial state to the final state, and
• in a level above the lowest level of the hierarchy, for each of a plurality of groups of adjacent sections, using weighting values calculated at a lower level of the hierarchy to calculate third weighting values indicating respective likelihoods that a sequence of data values comprising a group of adjacent sections corresponds to each of a plurality of possible state sequences, the group of sections comprising one or more possible initial states and one or more possible final states;
• and, for each said possible initial state and said each possible final state for each group of sections, selecting a third weighting value of highest likelihood corresponding to a state sequence from the initial state to the final state;
• and
• outputting information specifying a state sequence of highest likelihood according to the weighting value selections made during the hierarchical processing.