Novel Architecture And Efficient Algorithms Of Polar Coding And Decoding

In recent years,the fifth generation mobile communication(5G)technology has been born to meet the growing social needs and to cope with various new application scenarios that will emerge in the future.3GPP has defined three major application scenarios for 5G: enhanced mobile broadband(e MBB),ultra-reliable low latency(u RLLC)and massive machine type of communication(m MTC),and the channel coding performance requirements of 5G systems vary for different application scenarios.Polar code was formally established as a short code coding scheme for the control channel in the enhanced mobile broadband(e MBB)scenario of the 5G standard at the 3GPP meeting in 2016 due to its good theoretical performance,deterministic coding structure,and low complexity.Based on this,this dissertation conducts a study of novel architecture and high-performance decoding algorithms for polar codes with respect to their BER and computational complexity performance.The relevant research is summarized as follows.A novel transmission scheme for PAC codes based on spatially coupling is proposed in Chapter 3 of this dissertation.The proposed transmission scheme couples the end-register states at the end of encoding of the preceding and following code blocks,i.e.,the end-register state of the preceding code block is used as the initial register state of the following code block.Two methods of introducing zeroed convolutional code gain are proposed using the all-zero bits of the PAC code information side of the preorder,and the BER performance of the two gain methods is analyzed and compared.The numerical simulation results illustrate that under the condition that no additional complexity is introduced at the encoding side,the spatially coupled PAC code has a BER performance gain compared with the conventional PAC code,and the BER performance gain increases as the coupling depth increases and the list size of the list decoder increases.A soft information bit-flip decoding scheme based on improved key set is proposed in Chapter 4 of this dissertation.The decoding scheme cooperates a simple improvement of the key set by polarization weight ordering as the set of flipped bits in the subsequent decoding algorithm,and then the flipping criterion is formulated separately based on the two soft information matrices that pass the message in SCAN decoding,based on which three different SCAN-Flip decoding algorithms are proposed.Numerical simulation results verify the performance gain and computational complexity performance comparison brought by the proposed three decoding algorithms.The dissertation concludes with a stochastic computing-based SCAN decoding algorithm.Based on the low decoding complexity of stochastic computing and the small computational storage space required,this decoding algorithm uses stochastic computing in the SCAN decoding algorithm.We rewrite the node operation formula based on the determination of probability value calculation in SCAN decoding algorithm into the logic gate operation structure based on bit stream in r stochastic computing,and improve the performance of SC-SCAN decoding algorithm by increasing the length of random bit stream and re-randomization in the iterative process.