Method and apparatus for performing a generalized viterbi search to demodulate a sequence of symbols

An apparatus for performing a generalized Viterbi search method for maximum likelihood sequence estimation (MLSE) to demodulate a sequence of symbols, the apparatus comprising:

• a receiver front end for receiving the sequence of symbols;
• and
• a processor coupled to the receiver front end, the processor programmed to compute a metric corresponding to the sequence of symbols, the metric selected from one of:
• (a) a recursive metric having a structure
• m({overscore (a)}k)=m({overscore (a)}k−1)+{circumflex over (m)}(ak−N, . . . , ak−1, ak),
• wherein m({overscore (a)}k−1) and m({overscore (a)}k) are the cumulative metrics for symbols up to ak−1 and ak, respectively, and {circumflex over (m)}(ak−N, . . . , ak−1, ak) is the incremental metric, a function of (N+1) symbols {ak−N, . . . , ak−1, ak}, and
• (b) a non-recursive metric m({overscore (a)}k) that is an arbitrary function of a set of L submetrics {z1({overscore (a)}k), z2({overscore (a)}k), . . . , zL({overscore (a)}k)}
• m({overscore (a)}k)=m[z1({overscore (a)}k)z2({overscore (a)}k), . . . , zL({overscore (a)}k)],
• wherein the set of L submetrics have a generalized recursive structure
• z({overscore (a)}k)=z({overscore (a)}k−1)+{circumflex over (z)}({overscore (a)}k), l=1, 2, . . . , L;
• wherein the processor is further programmed to optimize the one of the recursive metric and the non-recursive metric by iteratively extending surviving paths on a trellis diagram to demodulate the sequence of symbols, wherein states of the trellis diagram at t=kT are specified by {ak−N′+1, . . . , ak}, and wherein N′≧1 is a predetermined integer selected to determine performance and efficiency of computation, and wherein 1≦N′≦N when the metric selected is the recursive metric.

A communication receiver for performing a generalized Viterbi search method for maximum likelihood sequence estimation (MLSE) to demodulate a sequence of symbols, the communication receiver comprising:

• a receiver front end for receiving the sequence of symbols;
• a user interface for interfacing with a user;
• and
• a processor coupled to the receiver front end and coupled to the user interface, the processor programmed to compute a metric corresponding to the sequence of symbols, the metric selected from one of:
• (a) a recursive metric having a structure
• m({overscore (a)}k)=m({overscore (a)}k−1)+{circumflex over (m)}(ak−N, . . . , a k−1, ak),
• wherein m({overscore (a)}k−1) and m({overscore (a)}k) are the cumulative metrics for symbols up to ak−1 and ak, respectively, and {circumflex over (m)}(ak−N, . . . , ak−1, ak) is the incremental metric, a function of (N+1) symbols {ak−N, . . . , ak−1, ak}, and
• (b) a non-recursive metric m({overscore (a)}k) that is an arbitrary function of a set of L submetrics {z1({overscore (a)}k), z2({overscore (a)}k), . . . , zL({overscore (a)}k)}
• m({overscore (a)}k)=m[z1({overscore (a)}k) z2({overscore (a)}k), . . . , zL({overscore (a)}k)],
• wherein the set of L submetrics have a generalized recursive structure
• z({overscore (a)}k)=z({overscore (a)}k−1)+{circumflex over (z)}({overscore (a)}k), l=1, 2, . . . , L;
• wherein the processor is further programmed to optimize the one of the recursive metric and the non-recursive metric by iteratively extending surviving paths on a trellis diagram to demodulate the sequence of symbols, wherein states of the trellis diagram at t=kT are specified by {ak−N′+1, . . . , ak}, and wherein N′≧1 is a predetermined integer selected to determine performance and efficiency of computation, and wherein 1≦N′≦N when the metric selected is the recursive metric.