Research On Low-Complexity Polar Code Decoding Algorithm

Polar code is a coding method based on the concept of channel polarization.It is the first one to be strictly proved to achieve channel capacity under Binary-Input Discrete Memoryless Channels(B-DMCs).The channel coding method,with a clear and simple coding and decoding algorithm,is regarded as a major breakthrough in coding theory.Polar code has been widely concerned by the academic community and has become a research hotspot in the field of channel coding due to its unique coding structure and reaching the Shannon limit.The successive cancellation(SC)decoding algorithm is essentially serial and requires higher real-time and lower complexity.Two improved decoding schemes are proposed by analyzing the construction and encoding and decoding algorithms of polar code.The main tasks are as follows:1.Channel polarization is studied and two stages of channel polarization are introduced: channel combining and channel splitting;three construction methods of polar code including the Bhattacharyya parameter method,the density evolution method and the Gaussian approximation method are studied.Meanwhile,the frame error rate(FER)performance of SC decoding under various construction methods is simulated.In addition,the principle of polar code encoding and decoding are studied,and the complexity of coding and decoding algorithms are analyzed.2.Aiming at the problem of high computational complexity of f-function nodes in SC decoding algorithm,the algorithm of polyline approximation operation is studied and applied to SC decoding algorithm,which is called polar code SC decoding algorithm based on polyline approximation operation.The algorithm approximates the hyperbolic tangent function and the inverse hyperbolic tangent function in the ffunction node to a 9-segment polyline function respectively,which reduces the computational complexity of the f-function node.The hyperbolic tangent function and the inverse hyperbolic tangent function in the f-function node can also be calculated by the quantization method.The hyperbolic tangent function and the inverse hyperbolic tangent function in the f-function node can also be calculated by the quantization method,the analysis results show that the improved algorithm can effectively reduce the computational complexity of the f-function node compared with the quantization method(when the quantization bit number is 5),and the proposed improved algorithmhas better frame error rate performance than the 5-bit quantization method.3.Aiming at the problem of low node utilization in SC decoding process,a SSC(Simplified Successive-Cancellation,SSC)decoding algorithm based on isolated information bit modification is proposed.The algorithm firstly modifies the isolated information bits in the SC decoding into frozen bits,and then uses the SSC decoding algorithm to simplify the calculation of the decoding tree.The proposed decoding algorithm greatly reduces the number of nodes calculated without sacrificing the performance of the FER,that is,reduces the computational complexity.The complexity analysis of the proposed decoding algorithm shows that the proposed algorithm reduces the computational complexity of SSC decoding by 7.79% when the code length N is 256 and the information bit length K is 128.