Novel low-latency low-power Viterbi decoder traceback memory architecture

In continuous-mode Viterbi decoding, the survivor path identification occurs continuously and does not wait until the complete message has been received. While this reduces the decoding latency time, there are problems determining the survivor path as the traceback starting state is unknown. The performance of 2 existing and 4 new methods of decoding on-the-fly are described and their performance analysed in terms of residual BER as a function of SNR and traceback memory length. A low-power traceback memory architecture is introduced. This is capable of performing a traceback in a single cycle, instead of the multiple read operations of a conventional memory. It can be used with any of the traceback algorithms described.