Efficient Polar Decoding Algorithms And Implementations For The Next Generation Mobile Communication Systems

As the unique codes that have been proved to achieve the symmetric capacity of the binary-input discrete memoryless channels,polar codes have been selected for the enhanced mobile broadband(e MBB)control channels in the 5th generation mobile communication systems(5G).The error correction performance of the successive cancellation(SC)decoding is not satisfactory with the finite code length.The performance-optimized version of SC is the cyclic redundancy check(CRC)aided SC list(CA-SCL)decoding.CA-SCL decoding has room for time complexity reduction with the unsuitable CRC distribution.Polar sphere decoding is an alternative maximum likelihood decoding on short codes,but the time complexity of its exhaustive algorithm is extremely high.List SD(LSD)reduces the searching space of SD.LSD has a huge error correction performance reduction on the low-rate and medium-rate codes.Four aspects of the improvements in polar decoding and construction are carried out: the SC decoding system,SD system,LSD system,and the decodertailored construction methods.In the SC decoding system,to reduce the decoding time complexity,CRC bits are put after the different information bits segments to create early-stopping criteria.To improve the error correction performance of the CA-SCL decoding,parts of the CRC bits are replaced with the BCH bits,and the CRC bits are allocated wisely according to the length and the reliability of the segmented bits.The combinations of SC decoding and stack and heap structures are proposed to reduce the decoding time complexity.In the SD system,to reduce the decoding time complexity,a synchronous determination method is proposed: the bits are arranged into several sets,where the Euclidean distances of the frozen bits could be calculated in time.Soft channel information aided SD algorithm is proposed.In the LSD system,the synchronous determination is applied to the LSD.Since the path evaluation is more precise,it is more possible to remain the correct path in the path selection on each decoding level.Therefore,the error correction performance of LSD on low-rate codes is improved(up to 2 d B).An efficient LSD based on path deletion is proposed to reduce the decoding complexity.a dynamic stack SD is proposed to reduce the complexity of LSD on high signal-to-noise ratios(SNR).A joint list decoding of SCL and LSD based on two threads is proposed to reduce the decoding latency.A fast-convergence genetic algorithm-based polar construction system is proposed to reduce the repeated calculation and the convergence period.An LSD-tailored new polar construction method is proposed.The performance of the LSD with the proposed construction is better than the other existed construction methods(exceed 2 d B on the low-rate codes).The time complexity of this construction is linear with the code length.In this paper,three kinds of polar decoding systems and the decoder-tailored polar code constructions are improved.The future research interests are the further algorithm optimizations and corresponding implementations.