Research On Technologies Of Network Fountain Codes And Their Applications In Wireless Sensor Networks

The amount of information required by people is increasing rapidly since wireless communication technology develops.How to transmit large scale data efficiently and reliably through limited network bandwidth has been the research focus in the field of wireless communication.Network coding allows relay nodes to encode the information,which can improve the bandwidth utilization,but network coding has large computational complexities.Digital fountain codes which have the advantages of low codec complexity and rate adaption,are very suitable for distributed networks.Network fountain codes,which combined digital fountain codes with network coding,can guarantee reliable information transmission and reduce the decoding complexity.Network fountain codes,which can balance network performance and computational complexity,have been a research hotspot in the field of network communication.This dissertation generates better network fountain codes based on fountain codes,the research work focuses on fountain codes mainly.In this dissertation,the study is focused on the network fountain codes,with emphasis on unequal error protection network fountain codes and the applications of network fountain codes in wireless sensor networks.The main work and contributions are as follows:(1)An unequal error protection(UEP)BATS(Batched Sparse)code with feedback is studied.BATS code is a kind of network fountain codes.To solve the problem that the less important packets of traditional expanding window BATS codes have high decoding overhead,a UEP BATS code with feedback is proposed.When the more important packets are decoded completely,an intermediate feedback message will be transmitted over the feedback channel to inform the encoder,the encoder adjusts encoding scheme,only select less important packets to create coded packets.Then,the bit error rate performance of UEP BATS code with feedback is analyzed via the And-Or tree evaluation,and the degree distribution for less important packets is optimized to further improve bit error rate performance of less important packets.Simulation results show that the proposed FEW-BATS code can decrease average overhead of successful decoding in less important packets and has no affect to decoding performance of more important packets.(2)A regularized variable-node degree BATS code,which regularizes variable-node degrees,is proposed to improve erasure floor performance of traditional BATS code.The bit error rate performance of regularized variable node BATS code is analyzed by And-Or tree evaluation,furthermore,check-node degree distribution of regularized variable-node degree BATS code is optimized to decrease decoding overhead.Simulation results show that regularized variable node degree BATS code can improve erasure floor performance and performance improvement regarding coding overhead could be introduced.(3)A UEP regularized variable-node degree BATS code is studied for wireless sensor networks(WSNs).By introducing regularized variable-node degree distribution to expanding window BATS codes,a UEP regularized variable-node degree BATS code is proposed to protect the packets with different priorities.The UEP performance of UEP regularized variable-node degree BATS code is analyzed,and check-node degree distributions are optimized to decrease decoding overhead.A novel network coding scheme based on the UEP regularized variable-node degree BATS code is designed for WSNs to transfer data hierarchically and improve erasure floor performance.To decrease average overhead of successful decoding in less important packets,an intermediate feedback message will be transmitted over feedback channel to inform the encoder that the more important packets are decoded completely,the encoder adjusts encoding scheme.Simulation results show that the UEP regularized variable-node degree BATS code decreases coding overhead and can improve erasure floor performance of packets with different priorities.(4)A novel network coding scheme based on online fountain codes is studied for WSNs.First,an enhanced online fountain coding scheme is proposed.An online fountain code is a kind of fountain cods with feedback.The optimal coding strategy is only set in the completion phase of online fountain codes,which can only achieve sub-optimal decoding performance.In enhanced online fountain codes,packets which have the minimum variable-node degree are selected to generate coded packets.Then,a novel network coding scheme based on enhanced online fountain code,namely enhanced online network fountain code,is designed for WSNs where multiple source nodes communicate with a single destination node through a relay node.The destination node transfer a feedback message expressed by a uni-partite graph to the relay node after the optimal degree is updated,then the relay node adjusts encoding scheme,packets which have the minimum variable-node degree are selected to generate coded packets.Furthermore,the upper bound on coding overhead of enhanced online network fountain code is derived.Simulation results show that the proposed enhanced online network fountain code can improve bit error rate performance and decrease average coding overhead,improvement of feedback overhead and coding complexity could also be introduced.