LORD: Tracking mobile clients in a real mesh
Résumé
Localizing a mobile client in a Wireless mesh network (WMN) consists in finding the router this client is attached to. Intuitively, the simplest manner to perform this task is to flood the whole network asking for the location of this client. Flooding the network is good for reliability but leads to increased latency and broadcast storm problems, affecting the efficiency of the localization service. In this paper, we design, implement, and evaluate LORD, a new client localization scheme based on a Distributed Hash Table (DHT). The originality of LORD is that it avoids exponential search on the DHT by relying on a proactive routing protocol running among the routers in the wireless backbone. Furthermore, it avoids flooding and provides fully transparency to end-users. Through experimental evaluation on a real WMN testbed, we show that LORD improves performance versus current flooding-based approaches by up to 3.5 times in terms of number of packets generated and 2.5 times in terms of link activations. Furthermore, although we mainly expected gains with reference to overhead, we also observed reductions of up to 50% in the handover time.