Research On Iterative Decoding Algorithms For Polar Codes

Nowadays,with the rapid development of wireless communication system,more advanced error correction coding technology is needed to improve the reliability of transmission process.In the selection of channel coding schemes for 5G(5thGeneration)enhanced mobile broadband scenario(eMBB),polar codes is chosen as the control channel coding scheme by the International Organization for Mobile Communications Standardization(ISO)3GPP.At present,as is one of the code scheme which can theoretically achieve the Shannon limit,to find efficient polar encoding and decoding schemes for achieving the better level of information transmission is the main research direction of polar codes.As a result,the main research work of the polar codes in this paper are as follows.First,the basic theory of polar codes is analyzed in detail.After analyzing the encoding and information bit selection schemes of polar codes,roundly performance tests and comparison of the basic polar decoding algorithms have been down,including the SC(Successive Cancellation),BP(Belief Propagation)and SCAN(Soft Cancellation)algorithm.Secondly,the decoding algorithm and hardware implementation of BP for polar codes is analysed in this paper.After finishing the analysis of the approximate algorithm which named the SMS(Scale-Min-Sum)algorithm,an improved BP algorithm named LUT-SMS(Look-Up-Table and Scale-Min-Sum)algorithm is proposed.This algorithm uses look-up function to approximate the non-linear function which used in the iteration formula of the original BP algorithm accurately,and it has been proved that the fitting effect of the look-up function to the original function is better than that of the SMS algorithm.The simulation results under AWGN(Additive White Gaussian Noise)channel show that the LUT-SMS algorithm that using the lookup-table function can achieve better performance than the SMS algorithm under long code length,and the maximum performance gain is about 0.3dB.At the same time,for the hardware implementation of SMS algorithm,an improved structure of SMS computing unit is proposed in this paper,which can reduce the computing cycle from four adder's latency to three while increasing only a small amount of resource consumption,thus achieving a throughput gain of about 32% in the case of full parallel decoding.Finally,by analyzing the improved algorithms of SCAN algorithm,which called RLSC(Reduced-Latency-Soft-Cancellation)and RLSC(Reduced-Latency-SoftCancellation)algorithms,an improved MRLSC(Modified-Reduced-Latency-SoftCancellation)algorithm is proposed in this paper to overcome the shortcomings of these two algorithms.With only a small increase in decoding delay,this algorithm eliminates the influence of the approximate iteration formula called MNA(Modified Node Activation)in RLSC algorithm by adjusting the position of information bits.The simulation results under AWGN channel show that the proposed MRLSC has the highest decoding performance gain of 0.4dB compared with the RLSC algorithm.