Research On Delay Tolerant Network Routing Optimization Based On Syntheticl Estimation Of Contact Metrics And Network Coding

Delay Tolerant Network (DTN) has been proposed to accomplish communications in peculiar network environment which cannot meet the requirements of traditional routing protocols that there has to be an end-to-end path at all times. Store-Carry-Forward is utilized in DTN routing for data transmission in order to overcome the shortcomings such as long transmission delay, high error rate and intermittent disruption. Therefore it has been extensively applied to deep space communication, satellite relay, military Adhoc network and sparse sensor network in remote regions. In recent years, as network applications spread and demands of service vary, employing network resource to the greatest extent has become the hot spot by the optimization of routing performance. This dissertation focus on how to find better schemes for data forwarding and message processing that improves DTN routing performance. The main innovations are listed as follows:(1)A routing protocol SECMR based on synthetical estimation of contact metrics is proposed in social DTN, by which the shortage that both contact metric with not enough precise estimation and information congestion in active node could be overcome. SECMR selects IPMM and RWP as the mobile model of nodes in social DTN and takes activeness of node, time length of connection and disconnection, frequency of contact and similarity of neighbors into account when making estimation of contact metrics so as to provide more accurate estimation model. Moreover, intra-region forwarding constraint factor SOC_CST is defined to limit the forwarding ability of intra-region nodes avoiding information congestion. Compared with Prophet and MEED, it is shown by simulation that SECMR could acquire better performance in social DTN that delivery ratio increases, average transmission delay decreases and information congestion eases.(2)Traditional network coding scheme needs too much coding nodes to improve the throughput and transmission efficiency of DTN routing. Thus, a network coding optimization GAMUCS based on mixed and uniform chaotic sequence and improved genetic algorithm is proposed. One-dimensional Logistic and Tent mapping is mixed and uniformed to generate initial population avoiding excessive dependence to the individual sequences. At the same time, the dissertation presents decentralized evolution, chromosome proving and adaptive adjusts of both cross and mutation probability. It is shown by simulation that GAMUCS could output the network coding scheme with fewest coding nodes and possesses faster convergence compared with GAS, MCN and IGAS. Moreover, as intermittent contacts increase, GAMUCS could improve throughput and reduce transmission delay more effectively than routing and entire NC.(3)A network coding scheme DNC-ER based on destination is proposed to improve the performance of epidemic routing in DTN, which overcomes the shortcomings that traditional network coding cannot fit for the environment when multi-flows are transmitting simultaneously. Messages with the same destination are encoded in DNC-ER in order to reduce the coding complexity and the waiting time before decoding in destination nodes. Furthermore, buffer management in DNC-ER decides whether to encode again or drop the buffered message by the destination. It is proved by theoretical derivation that our buffer management could decrease the types of messages in GBD and complexity of utility calculation and more precise utility calculation model is established simultaneously. Simulation results show that DNC-ER acquires striking promotion in both delivery ratio and average transmission delay compared with ER and NC-ER.