A novel SISO trellis strategy for relaying distorted signals in wireless networks

We consider the relaying of binary antipodal signals across two hops via soft regeneration and soft error protection. The signals at the input of the regenerator are degraded by additive white Gaussian noise (AWGN). Traditional approaches either directly relay the noisy analog waveform to the destination (thus missing the opportunity for coding gain), or make hard detection and then re-encode the bits using a digital error correction code (which may risk error propagation). This paper proposes a new class of soft-input soft-output (SISO) encoding strategies that are shown to outperform the conventional hard and soft forwarding schemes. Making essential use of the trellis structure, the soft encoder and the maximum likelihood decoder both operate efficiently in linear time, and support variable block sizes and code rates. Practical applications in relay communication are discussed. Simulations confirm the outstanding performance of the proposed scheme.