Bit-reliability Based Non-binary LDPC Decoding And Energy Consumption In Engineering Implementation

Low-density Parity-check Code(LDPC)is a capacity-approaching channel code and was LDPC code was agreed as the channel coding scheme for 5G eMBB data channel at the 3GPP meetings in 2016.Non-binary LDPC code is generalized from binary LDPC code.Compared to its binary counterpart,the non-binary LDPC code has many advantages in decoding performances,modem and channel interleaving and has been one of the hotspots in coding field.This paper focuses on the design of bit-reliability based non-binary LDPC decoding algorithms and their energy consumption analysis in engineering implementation.The main work and contribution are as follows:1.An enhanced weighted bit-reliability based(EwBRB)decoding algorithm is presented for non-binary LDPC code.For the presented algorithm,the Hamming distance coefficient is optimized by the column weight information.During the iterative decoding,the standard finite symbol used to compare with the extrinsic information is adjusted adaptively based on the plurality-logic and optimized corresponding to the information distribution along the variable nodes.Simulation results show that compared with the original wBRB decoding algorithm,the presented EwBRB algorithm can achieve a lower error floor with an acceptable increasing computational complexity,which can find application in the communication environment with extremely constraint of error floor.2.A new energy-based comprehensive evaluation of computational complexity is presented based on energy consumption in engineering implementation for decoding algorithms.For the new complexity analyzing scheme,the energy consumption is computed corresponding to the quantization parameter,the real/integer or logical operation and different operations(including the finite field addition,multiplication and comparison,etc.)among symbols from finite field construed in different orders.Compared with the traditional complexity analysis scheme(simply counting the operations),the presented scheme first converts different operations into energy consumption,so that the energy-based complexity can be compared in a uniform standard,which makes the complexity comparison become more fair and reliable.3.Based on the energy-based analysis scheme presented in this paper,we compare and analyze several typical MLGD and BRB non-binary LDPC decoding algorithms and the presented EwBRB decoding algorithm and get the energy consumption experimental data and for different decoding algorithms.The experimental results may provide some ideas in engineering implementation for non-binary LDPC decoding from the perspective of energy consumption.