Font Size: a A A

Polar Codes Based Joint Source-Channel Coding

Posted on:2020-08-28Degree:DoctorType:Dissertation
Country:ChinaCandidate:L Q JinFull Text:PDF
GTID:1368330575956363Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
Under the guidance of Shannon's source-channel separation theory,almost all of the existing communication systems are designed based on source coding and channel coding separation.The theory points out that when the code-length tends to infinity,separate source-channel coding(SSCC)achieves the performance of joint coding,but the infinite code-length also means infinite delay and complexity.In the case of finite code-length,joint source-channel coding(JSCC)outperforms SSCC.With the development of source coding and channel coding,their performances are very close to the information-theotical limits,the further gain need to be obtained by JSCC.Compared with the simple cascade of the optimal source coding and channel coding,JSCC achieves the overall optimization of the system through the interactive operation and joint design between source coding and channel coding.Polar codes,a kind of capacity-achievable codes,were recently proposed and attract the attention of many researchers.Polar codes in combination with JSCC,are used for the design of JSCC systems,which can effectively improve the performance of communication systems.Motivated by the fact that the application of polar codes in JSCC is vacant specifically for distributed sources,this paper will study and design polar codes based JSCC schemes for distributed sources.The main contributions of this work are summarized as follows:1.To solve the joint source-channel decoding problem,a distributed joint source-channel decoding scheme using systematic polar codes is proposed.In this scheme,the natural correlation among distributed sources is ultized at the receiver to improve the decoding performance and consequently save transmission power.In the proposed iterative decoding process,the log likelihood ratio(LLR)is modified with channel observations or hard decisions,which effectively solves the provlem that the typical decoder of polar codes isn't soft-output,and significantly improves the probability of correct decoding and the end-to-end system performance.The scheme is well-suited for multiple sources,non-binary sources,high-level modulation,and even the sources with unknown statistic properties.2.The joint source-channel coding problem with side information at the receiver is studied,and an optimal polar coding scheme is proposed.Based on the proposed quasi-uniform systematic polar codes,the scheme can jointly polarize the source with side information and the transmission channel.A quasi-uniform distribution is proposed to design the positions of systematic bits,which ensures the convergence of entropies of random variables after polar transform.In order to construct quasi-uniform systematic polar codes,bit-swap coding and efficient coding algorithm are proposed.The invertibility of submatrix of generation matrix is guaranteed by linear transformation of information bits,i.e.,the existence of systematic polar codes is guaranteed.The quasi-uniform systematic polar codes based Arikan kernels are extended to the general binary polar kernel.3.To solve the distributed joint source-channel coding problem in orthogonal channels,a distributed joint source-channel coding scheme is proposed.In order to adapt different code rates,punctured quasi-uniform systematic polar codes are proposed.The puncture pattern is also quasi-uniformly distributed,which ensures the convergence of entropies of random variables after polar transform.Due to the optimality of quasi-uniform systematic polar codes and punctured quasi-uniform systematic polar codes,the distributed joint source-channel coding scheme using two polar-like codes is also asyptotically optimal.4.For the problem of transmitting distributed sources over the multiple access channel,a polar coding scheme that can achieve the CES bound is proposed.The scheme is based on the mixed monotone chain rules,which jointly polarizes distributed sources and the multiple access channel.The reliable decoding is guaranteed by chain construbtion,and the distribution of codewords fed to the channel is guaranteed by randomized rounding.
Keywords/Search Tags:Distributed sources, independent channel, joint source-channel decoding, joint source-channel coding, multiple acces channel
PDF Full Text Request
Related items