Research On Efficient Fault Tolerance And Security Mechanisms Of Network Coding

Network coding is a novel data transmission approach, comparing with traditional routing methods which has been proven that can increase network throughput, improve load equilibrium, enhance robustness of network, be done in a distributed network with low complexity and potentially possess security superiority. Because of these advantageous characteristics, recently, network coding catches significant attention from more and more research communities.Aiming at the prominent problems of network performance (such as throughput, decoding complexity and energy overhead), security and fault tolerance (such as eavesdropping attack, error control and node selfishness) exist in previous network coding research schemes, based on the reasonable performance, the objective of this dissertation is to address the fault tolerance and security problems of network coding by combining the information theory with game theory. Major contributions of this dissertation can be summarized as follows:1) We propose a secure network coding scheme combining with incentive compatible mechanism, this scheme can solve the nodes selfish problem of network coding scenario and effectively prevent nodes'collusion attacks by applying game theory; meanwhile, it can resist the limited pollution and dropping attacks from macro perspective by utilizing corresponding encoding/decoding theory and adding the finite redundancy packets; moreover, it also decreases the compromised probability of the message and achieves well tradeoff between security and performance of network by integrating the multipath routing with the optimized coding packets allocation algorithm; finally, the simulation evaluation and security analysis confirm our theoretical results.2) We propose a compressed error and erasure correction scheme of network coding, this scheme reduces the computation overhead dramatically by employing efficient algorithm to detect and delete linearly dependent received packets in destination node; it also simplifies the hardware operations when scheme reduces the received matrix Y of destination node to form Ek (Y )instead of E (Y )in decoding process; when at most r original packets get combined, the author constructs a compressed code of coding vector to increase the payload of each packet from M - kto M - O ( r log qk), guarantee the received information can be accurately decoded (i.e., address error and erasure problems), and do not enhance the overall decoding complexity of the network system for the application of compressed code; finally, the numeric results and analyses illustrate the security and performance advantages of the proposed scheme.3) We propose a secret and reliable network coding scheme against a malicious adversary, this scheme ensures that the communicated information is complete reliable and information-theoretic security from the adversary when the network suffers from z pollution packets,μobservations andρpacket erasures; when the transmitted information is accurately and secretly decoded by destination node, the communication network can achieve a maximum transmission rate of k -μ- z-ρ(upper bound of this network scenario) under the requirement of 2 z +ρ+μ< k≤2 z+ 2ρ, and it does not require the packet length grow to infinity; finally, the numeric results and security analysis confirm the characteristics of the proposed scheme.4) We propose a secret and reliable coding scheme for noncoherent multisource network coding, this scheme addresses the problems which induced by z pollution packets,μobservations andρpacket erasures in multisource network coding scenario; when the transmitted information is accurately and secretly decoded by destination node, the communication network can achieve an asymptotic transmission rate of k - 2 z -ρ- 3μ, it's similar to the single-source scenario, this scheme also does not require the packet length grow to infinity; moreover, the intermediate nodes only perform random linear combination operations and are oblivious to the concrete encoding and decoding algorithms implement in network.