Computing the full path metric in viterbi decoding
AssigneeTexas Instruments Incorporated
InventorDale E. Hocevar
Life of this patent
Estimating the term…
Abstract
By utilizing an additional counter and monitoring the maximum state metric at each stage, only forward progressing modulo wrap-arounds will occur and these can be counted. After decoding this count information, it can be used with the initial and final state metric values from the decoder to compute the desired full path metric. The method only requires monitoring state metric wrap-arounds moving in one direction and hence only needs to increment the extra counter as opposed to having to do likewise in the opposite direction. In another embodiment, the method can handle both forward and backward progressions by incrementing and decrementing a counter.
Claims20
A method for computing the full path metric (FPM) in a Viterbi decoder that uses modulo normalization, comprising the steps of:
- (a) determining the maximum state metric value at each stage;
- (b) comparing the maximum state metric at each stage to the maximum state metric at the previous stage to determine if a marker point has been crossed;
- (c) incrementing a counter if the marker point has been crossed in step (b);
- and
- (d) determining the FPM using the formula:
- FPM=(M+Cadj)−I−d
- where D=(M−E)mod(K), and where “M” is the maximum state metric value for the last stage at the end of frame, “E” is the state metric value for the ending state of the frame used for traceback (e.g., the last state in the decoded path, and also the known ending state in the encoder), “I” is the state metric value for the initial state of the frame, K=2t is the size of a modulo ring where “t” is the number of bits used to represent the state metric values, and Cadj is the value of the counter when it is left shifted by “t” bits.
Dependencies are inferred from the claim text.