Improved SOVA-based decoding algorithm for tailbiting codes on wireless channels

Tailbiting convolutional codes will be used for several applications in new cellular mobile radio systems. This encoding method does not reset the encoder memory at the end of each data block, avoiding the overhead of the zero tail and improving the efficiency, especially when encoding short data blocks. Nevertheless, the absence of a known tail highly increases the complexity of the decoding process. This fact added to the high computational burden of maximum likelihood (ML) decoding have made necessary the development of decoding algorithms which achieve good performance in terms of computational complexity and error correction capabilities. In this work we propose a novel decoding algorithm for tailbiting codes which is suitable for wireless environments with steep variations in channel conditions. The proposed method performs two different Viterbi decodings of the received data. In the first one, the most likely state is estimated with a modified version of the soft-output Viterbi algorithm (SOVA). If the path with the best metric decoded in the first step is not tailbiting, a second Viterbi decoding is performed. In this case, the state estimated in the previous step is employed as the initial and final states of the trellis. Simulations results obtained in an orthogonal frequency division multiplexing (OFDM) system over a wireless channel are close to the performance of the maximum-likelihood decoding and other proposed algorithms.