| In recent years,the Internet of Things(IoT)technology based on Wireless Sensor Network(WSN)has developed rapidly.In order to meet the 5G requirements,and to realize the 6G vision,the diversification of business requirements,the 1000-fold increase of data flow,and the connection of billions of devices will pose challenges to the design of communication systems.Because of its low cost and low power consumption,Joint Source-Channel Coding(JSCC)technology can meet the core technical requirements of IoT and provide a strong support for the realization of the interconnection of all things.For joint source and channel coding,one has to joint design the source coding and channel coding to achieve the optimizing performance.So,in this paper,we try to analyze the matching criterion of different compositions in the transmission system based on JSCC and optimize the system.In this paper,the main innovations include:1)To improve the error-floor performance and robustness of the JSCC system,the design of source codes based on the matching criterion between source coding rate and source statistics is proposed.Firstly,it is found that optimized protograph LDPC codes for channel coding are not optimal for source compression,which indicates that the source codes in the JSCC systems need to be designed specifically.Secondly,by using the Protograph Extrinsic Information Transfer(PEXIT)algorithm for source codes and the Differential Evolution(DE)algorithm,source codes with high source decoding thresholds are designed.The PEXIT analysis and the BER simulation have shown that the proposed source codes obtain higher source decoding thresholds than conventional codes,which extends the range of compressible entropies.So,it can be a constructive way to reduce dependence of this system on source statistics,especially in high SNR region.What’s more,the proposed codes help to obtain a lower error floor than conventional codes.2)To improve the waterfall performance of the JSCC system,the matching criterion between source code and channel code in the JSCC system based on protograph LDPC is investigated and used to deal with the joint optimization problem of source codes and channel codes.Firstly,we extend and modify the curve-fitting algorithm for channel codes and propose a matching criterion of source codes and channel codes.As the criterion can not be applied in code searching directly,the cost function to measure the width of decoding tunnel is constructed,and a joint optimization procedure based on Geneitic Algorithm(GA)is proposed to search for the optimal source code and channel code with the lowest decoding threshold for(asymptotically)long and short-to-moderate-length codewords.The decoding thresholds and simulation results show that the optimized source codes and channel codes can achieve remarkable gains on the decoding threshold than conventional(benchmark)codes.Also,the optimized codes can obtain better BER performance both for(asymptotically)long codewords and short-to-moderate-length codewords.3)To improve the performance of the transmission system based on JSCC,the matching criterion between JSCC and coding channel is investigated.According to the matching criterion,a novel matching design of the transmission system is proposed by using the redundant information of source coding,which can obtain the probability amplitude shaping gain of the coding channel.Firstly,the Bit-Interleaved Joint Source Channel Coded Modulation(BIJSCCM)system is proposed to optimize the overall performance of the transmission system.The matching criterion between JSCC and coding channel is established by using the bit-level probabilistic amplitude shaping algorithm.Secondly,the code parameter and interleaving are designed based on this criterion.So that the residual redundant information can not only be utilized by the decoder,but also be used to obtain the probability amplitude shaping gain.In addition,DE algorithm is used to search the source and channel codes with excellent performance.Theoretical analysis and simulation results show that the redundant information of source coding in the BIJSCCM system can not only be used to improve the decoding performance,but also obtain the probability amplitude shaping gain of the coding channel. |
