Towards scalable routing for wireless multi-hop networks

Abstract

Infrastructure-less wireless multi-hop networks have long been proposed for natural disaster and warfare scenarios. However, the current demand of such networks has been towards social networking, gaming and ultimately, ubiquitous computing. In fact, the increasing number of users that own wireless capable devices is taking these networks to an entirely different scale. Existing routing protocols do not scale and do not consider the context wherein services operate. By presenting an alternative routing protocol that appropriately handles mobility of users among different contexts, large-scale clustered wireless networks are designed, using an efficient gateway selection with load-balancing capabilities. This protocol is based on the PhD work that defines a routing approach which uses a virtual hierarchy of clusters to explore the contextual-proximity of nodes, while reducing the total overhead of routing traffic even when compared with other cluster-based approaches. Moreover, this protocol also takes advantage of a Kernel-based link quality estimator, also considered during the course of the PhD thesis, which is capable of predicting gateway link disconnections, increasing the total amount of delivered data. The obtained results reveal that this routing scheme outperforms existing routing protocols regardless of the mobility pattern being used, being consistently lighter in overhead and delivering up to 50% more data traffic. These results motivate a new era of large-scale wireless multi-hop networks suitable for hand-held devices exchanging data amongst themselves.