Redundancy-based Approaches in Wireless Multihop Network Design.
详细信息
文摘
Wireless multihop networks such as ad hoc and mesh networks consist of wireless nodes without any fixed base stations or a backbone infrastructure. Such networks are characterized by frequent host mobility,limited power reserve and interfering transmissions. The wireless medium is susceptible to random uncertainty of radio links as well as malicious jamming. While the former is observed in the form of intermittent link failures,the latter is observed as geographically correlated failures in the network. Routing in such networks usually needs to be robust,so as to adapt itself to the dynamics of the network. Traditional unipath routing techniques reduce the overall packet delivery ratio in presence of network failures. Multipath routing is a common approach to increase the reliability and end-to-end throughput. In this research,we first explore some redundancy based multipath routing techniques in wireless mesh networks. We consider a network of nodes addressed by their locations,and propose a diffuse pathset routing technique. The goal of this approach is to increase reliability of transmissions in failure prone networks. The geo-diffuse multipath routing technique,called Petal Routing,takes advantage of the broadcast nature of wireless networks to reduce the number of transmissions over multiple paths. We compare this approach to a geo-diverse routing technique,which carries out power control to protect against jammers in the network. We highlight scenarios under which one performs better than the other,and also compare them to existing multipath routing techniques. Transmissions in wireless networks may require performance guarantees,which can be achieved by using advanced routing strategies. Such guarantees may not be delivered due to vulnerabilities such as incorrect neighborhood information,high probability of random link failures and jammers. We examine one such performance metric,namely reliability,or packet delivery ratio. Being able to predict the reliability of a transmission a priori can be useful in designing robust networks. To this end,we present an analytical model that can predict the reliability of a transmission being carried out using Petal Routing,given certain parameters such as network node density and region of network involved in the transmission. We also present a more general predictive model for wireless networks inspired by hybrid control theory. We model the network as a stochastic dynamical system where the wireless links are either stationary availability probabilities or Markov transition probabilities. The system we present has a centralized controller that pre-computes optimal paths to the destination by carrying out reachability analysis in order to obtain predictive results. We investigate the accuracy of the theoretical model by comparing reliability predicted by the model and that achieved by OPNET simulations of standard routing protocols such as OLSR and AODV. This technique can provide the highest achievable reliability for a given transmission,and thus be used for benchmarking and improving existing routing protocols to make them more robust to wireless failures. Lastly,we explore the timescale of wireless link failures in an effort to make our predictive model more realistic. While the two-state up/down Markov model can be used to represent wireless links,the actual transition probabilities from these states can vary widely based on what the observation timescale is. To make our model emulate real wireless links,we carry out some experiments on 802.11 wireless links and measure the timescale of link level fluctuations. The results of these measurements can be used both to design realistic predictive models,as well as routing protocols that consider the link level fluctuations while deciding on the best next hop to deliver a packet to.