Research On Sequential Decoding Algorithms Of Spinal Codes

With the rapid development and wide application of wireless communication technology,people have higher demands for efficient and reliable transmission of information.One of the key problems in wireless communication systems is how to implement a communication protocol with high throughput under time-varying channel conditions.Different from the fixed rate coding scheme,rateless codes can adaptively alter the number of coded symbols transmitted according to the interference,without estimating the channel conditions or adjusting the coding rate explicitly.Spinal codes,a new type of rateless codes,have a simple coding structure and can be proved in theory that it can achieve the Shannon capacity in both binary symmetric channel(BSC)and additive white Gaussian noise(AWGN)channel.The sequence decoding algorithms of Spinal codes are investigated in this thesis.In order to reduce the decoding time and improve the decoding efficiency,an unquantized Fano decoding algorithm with a limit on the backtracking distance is proposed in this thesis.In this algorithm,the decoder can only move back in the coding tree for limited distance.The threshold of path metric will be reduced when the decoder reaches the limit of backtracking distance,so that the decoder can avoid moving back and forth repeatedly in the coding tree.As a result,the decoding algorithm can converge as soon as possible,and the decoding results can be obtained in a shorter time.Compared with the modified unquantized Fano sequential decoding algorithm?Bubble decoding algorithm and the forward stack decoding algorithm,the new decoding algorithm improves the decoding efficiency without sacrificing rate performance.In addition,based on the forward stack decoding algorithm,this thesis proposes a dynamic layered forward stack decoding algorithm that can adjust the parameter of the coding tree adaptively according to the channel conditions.The size of a layer can be dynamically adjusted according to the metric information of the nodes in the storage space in this algorithm,so as to provide better resilience to noise when the channel states vary.The proposed algorithm has lower computational complexity and higher throughput than the forward stack decoding algorithm.Finally,in order to improve the rate performance of Spinal codes,this thesis designs a segmented encoding scheme of Spinal codes,based on that the parallel unquantized Fano decoding algorithm is proposed.After divided into several segments,the original information bits are encoded independently,and the code symbols generated from different encoders are transmitted as a whole frame.The new scheme combines the outstanding complexity performance of short code length and the transmission efficiency of long code length at the same time.During the decoding process,more than one decoder tries to build the code tree independently from the initial state of each segment.Simulation results show that,the parallel unquantized Fano decoding algorithm with a segmented encoding scheme can achieve higher rate performance with lower complexity.