A protocol for robust and efficient service discovery in large, highly mobile radio networks

Abstract

Existing service discovery mechanisms for ad hoc networks are often designed with one specific network type in mind. Solutions capable of handling highly mobile nodes usually have high bandwidth requirements, particularly as the number of nodes increases. The bandwidth requirement can be reduced by locally caching state information, but this increases the risk of nodes having outdated state information when mobility is high. Some solutions avoid these two issues by tightly coupling service discovery with the routing mechanism itself. However, this requires that nodes are homogeneous on the network layer. We propose a solution that leverages the special properties inherent in broadcast-based radio networks. In such networks, every node within transmission range will hear a transmission, be it unicast or broadcast. Each node therefore aggregates relevant service information and broadcasts it at regular intervals. Unnecessary transmissions are suppressed by efficiently synchronizing local state information. In this paper, we describe the Mist-protocol, a robust and efficient adaptive service discovery protocol, that supports large, highly mobile networks consisting of heterogeneous nodes. We test the protocol in large scale simulations in both static and mobile environments. Finally, we show that it is feasible to actually implement the design by providing a proof-of-concept prototype, which has been evaluated in a small scale experiment.