A new survivor memory management method in Viterbi decoders: trace-delete method and its implementation
暂无分享,去创建一个
The well known methods for survivor path storage and decoding are the register-exchange method (REM) and the trace-back method (TBM). The REM is conceptually simple, but it is not appropriate for VLSI implementation because it requires large power consumption and large chip area. The TBM is the preferred method in the VLSI implementation of Viterbi decoders (VD) having large constraint length and high performance. However, the TBM requires last-in-first-out (LIFO) buffer and has to use multiple read operations for high speed operation. This multiple operation results in complex control logic. In this paper, we propose a new survivor memory management method called trace-delete method (TDM) and realize this algorithm in hardware (H/W) for VLSI implementation and we compare the TDM with the TBM in terms of latency, the number of memory elements, and the requirements of control logic. The main advantage of the proposed method can be found as short latency and less requirements on additional control logic. Especially, if we combine the TDM with block interleaving the implementation is even simpler than the TBM. The method is studied with particular relevance to HDTV.
[1] R. E. Peile,et al. Stanford Telecom VLSI design of a convolutional decoder , 1989, IEEE Military Communications Conference, 'Bridging the Gap. Interoperability, Survivability, Security'.
[2] C. Rader. Memory Management in a Viterbi Decoder , 1981, IEEE Trans. Commun..
[3] P. Glenn Gulak,et al. Architectural tradeoffs for survivor sequence memory management in Viterbi decoders , 1993, IEEE Trans. Commun..