Research On Convolutional Network Coding And Decoding Algorithms For Variable Channels

With the rapid development of mobile communication,Internet services,and streaming media transmission occupying people’s lives.People are increasingly dependent on wireless communication equipment and have higher and higher requirements for wireless communication quality.The prior work mainly adopted automatic repeat request(ARQ)and forward error correction(FEC)coding schemes to handle the problem of wireless channel packet loss.Nowadays,5G device-to-device(D2D)communication technology has many advantages.However,without the support of base stations,its anti-interference ability is extremely poor,especially in long-distance transmission,there will be a high packet loss rate,which requires other means to improve transmission reliability.Network coding is an information exchange technology that integrates coding and routing.At the same time,the unique construction form of convolutional codes can fully utilize the advantages of network coding technology.This thesis mainly considers the simulation on the error uniform distribution channel(Bernoulli channel)and the burst error GE channel,which can better simulate the wireless network channel.This thesis studies convolutional network coding with a particular coding structure to problems in different channels.The specific contents include the following:Firstly,a truncated convolutional network coding algorithm based on convolution and ARQ retransmission mechanism is proposed in the single-hop point-to-point network model.The algorithm uses a particular coding structure in the source node,that is,sparse convolutional coding is carried out for the block code,and decoding is carried out at the sink node while receiving.The decoding time is recorded if the decoding conditions are met.Otherwise,joint decoding is carried out when the next group of data packets is sent.When the convolution depth of is reached if the receiving matrix of the sink node cannot be decoded correctly,the feedback information is sent to the source node,and a coded retransmission packet is generated for retransmission operation.The algorithm is simulated and compared under the Bernoulli and Ge channel models.Due to its particular coding structure and the characteristics of receiving while translating,it can achieve the throughput of RLNC large packets and has the delay of RLNC small packets,which proves that the algorithm has obvious advantages.Secondly,a random convolution network coding transmission scheme is proposed.The generation matrix is obtained by a cyclic shift of the basic generation matrix,and the basic generation matrix is generated by RLNC.The improved transmission scheme realizes the feedback of received information while transmitting and decoding.With this transmission method,the correlation between data packets is closer,and the feedback retransmission is more timely.In addition,the coding parameters are analyzed theoretically by the zoom method.The transmission scheme is simulated on the Bernoulli channel and GE channel and compared with the simulation results of truncated convolutional network coding and RLNC to verify the rationality and superiority of the transmission scheme.At the same time,the influence of adding unequal redundant data packets to ensure reliability on network parameters is simulated on the GE channel,which lays the foundation for adding unequal redundancy in different application scenarios.Finally,based on the truncated convolutional network coding transmission scheme,a transmission scheme of adaptively inserting unequal redundant packets is proposed.The scheme adopts a two-stage forward error correction mechanism.Under the framework of convolution structure,the algorithm sends partial redundancy according to the high bit rate in advance and then adaptively sends redundant data packets according to whether the feedback can be decoded.And when the convolutional structure is used,the latter redundancy will also help correct the former errors to achieve high-speed and low-latency transmission.The transmission scheme is simulated and analyzed under the Bernoulli and Ge channel models.Compared with the case of sending data packets at a fixed bit rate,the convolutional network coding algorithm with adaptive insertion of unequal redundant data packets has better delay performance.