Research On Low Complexity Encoding And Decoding Scheme Of Spinal Code

With the rapid development of cutting-edge technology in recent years,the challenges faced by wireless communication technologies are increasing in complex communication scenarios.The conditions of the transmission channel change over time,but the traditional channel encoding techniques are often difficult to meet the needs of some modern communication scenarios.In order to counter a series of real-time changing noise interference in the wireless channel environment,secure wireless communication system for efficient and reliable information transmission under complex time-varying channel conditions,researching encoding techniques with higher gains is always a hot trend in the field of wireless communication.This thesis focuses on an emerging encoding technology which named Spinal code for detailed discussion and research.The Spinal code is a new type of rateless encoding technique discovered in recent years.The idea of the Spinal code is to use a structure similar to the convolutional code to randomly encode the input bit sequence by introducing a hash function.The encoded codewords generated by such encoding structure have excellent pseudo-random characteristics,so that the Spinal code has better anti-interference capability under low SNR conditions.Its rateless characteristic makes it adaptable for dynamically channel,and it can control the code rate by itself,finally achieve the reliable communication of information;but in special application scenarios such as computing,storage,energy consumption and other limited environments,the Spinal code has high complexity due to its code structure.How to reduce the complexity of the Spinal code and improve its overall performance is the focus of this thesis.The previous part of this thesis describes the structure of the Spinal code,discusses the specific form of its encoding and decoding structure deeply,then we improve the encoding transmission mode and decoding scheme of the Spinal code.The details are as follows:(1)For the purpose of reducing the decoding calculation of the Spinal code,this thesis proposes a conception named as "fast restart" decoding.Considering the decoding process of the current Spinal code lacks an state determination mechanism,the decoding operation unable to terminate early when the decoding tree does not have the correct candidate path,which cause frequently redundant expansion of the decoding tree.To avoid the expansion of redundant nodes,The "fast restart" decoding scheme uses BP neural network to analyze the state of the decoding tree,and determines the after decoding process of the Spinal code according to the analysis result.The research results show that compared with the original decoding mode,the Spinal code can achieve lower decoding computation with almost no loss of code rate performance by using the "fast restart" decoding scheme.(2)For the problem that the long code complexity of the Spinal code is too high,we propose a "segmented CRC-aided" scheme for the Spinal code.Due to the Spinal code is decoded by extending the decoding tree,while the codes have a long length,the size of the decoding tree also increases sharply,which ultimately leads to a high decoding complexity of the Spinal code.And because the information content transmitted by the Spinal code is related to the feedback result of the last decoding,simply short code transmission will make the Spinal code need to maintain a large amount of decoding feedback.To solve the above problems,the "segmented CRC-aided" scheme refines the encoding information,it divides a high expansion large decoding tree into multiple low expansion small decoding trees.Utilize the characteristics of the Spinal code that long code transmission require fewer feedbacks and short code transmission has lower decoding calculation,which makes the Spinal code perform better with fewer feedbacks.The research results show that under the condition of low SNR,the "segmented CRC-aided" scheme can effectively reduce the decoding calculation of the Spinal code.