Transmission Performance Analysis And Optimization For BATS Code

In wireless multi-hop network,a source node would like to reliably transmit data to a destination node via multiple relay nodes.However,wireless communications experience severe packet loss due to the multipath effect and congestion.The more the number of hops is,the higher the packet loss probability.To provide the end-to-end reliability in multi-hop network,various techniques such as retransmission,network coding and fountain codes,have been proposed to resolve the packet loss issue.However,these techniques are not efficient for multi-hop networks due to high complexity,low throughput and long delay.As a novel channel coding scheme,BATS code has the unique advantages of lower coding and decoding complexity,which only requires limited space to complete coding and decoding,providing solutions for many scenarios in multi-hop network.In wireless multi-hop network,it is of great significance to optimize BATS code to improve the transmission performance,adopted as the channel coding scheme.The main contributions of this thesis are as follows:1.The normalized rank of batches of BATS codes in the time-variant channel is studied,revealing the maximum achievable rate for each packet.Packet loss rate on the link is described as a random variable following finite interval normal distribution,and then the normalized rank of BATS codes in the time-variant channel is studied and expressions of normalized rank of BATS codes based on random linear network recoding and systematic recoding are derived,which are verified by simulations.2.The properties of elements on diagonal of rank transfer matrix of BATS codes are studied.Based on the researches,an infinite-length optimal model of BATS codes is proposed.This model,where the sink node is cache free,estimates the degree range of batches of BATS codes.It can improve the transmission process by sending a feedback information after receiving a certain number of batches.When the source node adopts this model to transmit packets and the value of degree of batches locates in the estimated range,it can reduce the number of batches needed to transfer and improve the transmission performance significantly.3.The encoding process,decoding process and advantages of LDPC codes are studied.And an improved transmission model of BATS codes based on LDPC-precoding is proposed.LDPC codes are adopted as a precoding process of BATS codes,making best of capacity of error correcting of LDPC codes.All precoding words can be recovered via the check matrix of LDPC codes after receiving a certain amount of precoding words.Meanwhile,a joint BP decoding algorithm combining BP decoding of LDPC codes with BP decoding of BATS codes is proposed.This proposed algorithm is capable of recover more codewords from decoded LDPC codewords to reduce the number of batches and improve the transmission performance significantly.