Log-likelihood ratio algorithm for rate compatible modulation

Abstract

Seamless Rate Adaptation(SRA) based on rate compatible modulation(RCM) is a new receiver rate adaptive method, in which a block of bits are mapped into a series of symbols by a sparse matrix for high order modulation. According to channel state, the receiver select of the proper number of symbols to iterative demodulation by RCM. This system can achieves smooth rate adjustment under highly dynamic channel conditions. However, for fast implementation, the convolution in the horizontal processing and the vertical processing required lots of multipliers in RCM demodulation algorithm using Belief Propagation(BP), which beyond the current hardware. Therefore, it is difficult to make full use of the parallelizable of the BP. After analysised the principle of the RCM, by the proper deformation, we have proposed the LLR-RCM algorithm, in which we used the log likelihood ratio (LLR) to iterative demodulation. So the convolution in the horizontal processing are transformed into additions and a function table, all the multiplications in the vertical processing are converted to additions, what's more, it save the normalization and make the possibility for fast implementation by hardware.