Research On Decoding Algorithm For Polar Codes And Its Application In Image Transmission

With the rapid development of the information age,people have increasingly high requirements on the reliability of information transmission.Choosing channel coding with better performance is the urgent problem to be solved in the field of channel coding.In 2009,E.Arikan proposed a new channel coding method,called the polarization code.Polar code is the only good code that can achieve the Shannon's capacity limit up to now,and has a low complexity of coding and decoding.Therefore,it is widely used in the field of error correction coding.The core of polar code is the channel polarization.After channel polarization,the reliability of each polarised channel is different,and information bits and frozen bits are transmitted according to the difference of sub channels.Although the theoretical performance of the polar code can achieve the Shannon's capacity limit,but for the medium and short code length polar code,the performance of successive cancellationthe decoding algorithms is seriously lost due to incomplete channel polarization,so we need better decoding algorithm.In view of the error propagation characteristic existing in SC decoding process,this thesis proposes an improved successive cancellation bit-flipping decoding algorithm.The new algorithm uses CRC checksum and bit-flip to detect and correct error of the result of SC decoding.In order to correct the error bit in time and curb the error propagation better,the proposed algorithm segments the information sequence in the coding phase,and adds CRC check bits in each segment.In the decoding phase,bit-flipping was carried out for each failed CRC-check segment at the end of each segment decoding process.In addition,this thesis gives two strategies: uniform segmentation and self-heuristic segmentation.Self-heuristic segmentation can obtain better decoding performance by exploiting the channel polarization characteristics.Compared with the SC decoding algorithm,the ISCBF algorithm can obtain about1.19 dB SNR gain.On the other hand,with the increase of communication speed and processing capacity,image transmission has become an important form of communication which is commom in real life.The reliability of each polarization channel sub channels are different after channel polarization,and then transmit information bits with high reliability channel,so the polar code itself has unequal error protection(UEP)characteristic,this characteristic is used to the image transmission can effectively reduce the distortion rate in image transmission.In this thesis,the UEP coding scheme for the original image is designed firstly,mapping the high bit in every pixel of the image to the sub channel with high reliability,and mapping the low bit to the sub channel with low reliability.Then,according to the embedded zerotree wavelet(EZW)characteristics that the importance of the compressed bit stream are decreasing,designed UEP encoding scheme for compressed image.Using two UEP mapping schemes,the first oneis based on rate allocation,The compressed code stream is segment by rate adaptation,and uses non decreasing code rate to transmit each code stream.The second one is based on polarization characteristic which uniformly segments the compressed bit stream,then maps the forward bit stream to the high reliability channel of each frame,and maps the post bit stream to the low reliability channel of each frame.The simulation results show that when the SNR is 2dB and the ISCBF decoding algorithm is used,for the original image,UEP scheme can achieve 7.51 dB PSNR gain compared with non-UEP scheme.For the compressed image,compared with non-UEP scheme,the UEP scheme based on polarization characteristic can obtain 8.02 dB gain and 4.59 dB PSNR gain using UEP scheme based on code rate allocation.At the same time,compared with the SC decoding algorithm,ISCBF decoding algorithm can obtain about 7.53 dB,4.13 dB and 4.04 dB PSNR gains respectively.