Bio-inspired Multipath Optimization Transmission Model And Scheme Study

With the rapid increase in Internet applications and the continuous emergence of new services, network traffic grows greatly. Also, due to the lack of effective management and control, growing imbalance about Internet traffic tends to cause network congestion, which is becoming an important factor about reducing network performance. Nevertheless, multipath transmission is a technique to balance the network traffic, enhance transmission robustness and improve transmission efficiency, which is attracting more and more attention of researchers. Used some biomathematical models for reference, for example, adaptive attratcor selection model that models metabolism behaviors of Escherichia coli in changing nutrition environments, and biological competition model (Lotka-Volterra model) that describes dynamic changes of species population in an ecosystem, the multipath transmission optimization is studied further in WSN (wireless sensor networks), fixed networks, and multihomed networks. And, the main work is as follows:(1) A path selection model is proposed, which is inspired by the Escherichia Coli’s metabolism behaviors. In multipath transmission of WSN, greedy path selection is prone to cause frequent changes of the working path between each source and destination pair, and the frequent changes always lead to unpredictable network congestion and packet reordering. According to analogy between path transmission quality and mRNA concentration, the adaptive path selection process in the WSN multipath transmission is modeled by redesign multi-attractor equations. In the proposed model, an adaptation goodness formula is put forward in dynamic WSN network environments, which is defined as path-activity. Furthermore, to concretely specify the stochastic effect of the noise on the path selection, a novel analytic expression of the attractors containing the noise is first deduced in the multi-attractor equations. By modifying a single path protocol AODV (ad hoc on-demand distance vector routing) in OPNET simulator to be a multipath AODV protocol, path selection simulation scenario is setup. Then, to obtain the path transmission quality and path-activity, a path quality probe scheme is designed in the multipath AODV protocol. Finally, in five network simulation scenarios based on the multipath routing, we compare the performance of the proposed model and the greedy path selection on reducing the average network delay and the path oscillation. Through expressing the path quality as path end-to-end delay and maximum node path arrival rate of each path in specified WSN scenarios, simulation results present the proposed model can perform better than greedy path selection about reducing average network delay and path oscillation rate.(2) An OSPF (open shortest path first) path selection scheme is proposed based on the adaptive attractor selection model. Current IP routers only support equal cost multipath routing, which also only performs random path selection or traffic uniform distribution among equal-cost paths. Therefore, an adaptive path selection scheme is presented based adaptive attractor selection model, which can dynamically select transmission path according to realtime path quality. In the proposed scheme, a path selector is designed in IP (Internet protocol) network layer, which can perform realtime path selection of the proposed scheme, concretely for example, realtime path quality probe and routing update. By embedding the path selector into the OSPF protocol of OPNET simulator, IP simulation scenario is setup based FTP (file transfer protocol) services. Compared with greedy path selection scheme and uniform random path selection scheme in configured IP scenarios, simulation results show the proposed scheme gains better performance on reducing file transmission time, traffic throughput and traffic dropped.(3) An adaptive CMT (concurrent multipath transfer) traffic distribution scheme is proposed based on the Lotka-Volterra competition model. The scheme can optimize traffic distribution in multipath transmission of multihomed networks. In multihomed networks, two competition modes are concluded. One is multiple S-D (source-destination) streams competition for one path, the other is multipath competition for traffic between each multihomed S-D host pair. According to these two competition modes, two network competition models are redesigned based the Lotka-Volterra model. Subsequently, to apply the models into multihomed network system, a traffic distributor is designed, which is used to embed the traffic distribution schemes into the multihomed networks. Based on the traffic distributor, the paper compares the proposed scheme in OPNET simulator with the other three schemes, which is uniform traffic distribution scheme, greedy path selection scheme and uniform random path selection scheme. Simulation results show the proposed scheme can perform better about decreasing file transmission time and increasing network throughput in FTP service in configured multihomed network scenarios.This paper attempts to explore and exploit some study models and technical schemes from biology field to improve performance of multipath transmission. Through bio-inspired ideas, it is expected to relieve some difficulties of current network development, for example, dynamic, survivability and heterogeneous, and also, to provide a scientific reference for the evolution of the current network and the next generation Internet design.