Efficient event and query distribution in sensor networks

Abstract

A sensor network consists of a large number of sensors which are equipped with sensing, computation, and communication devices. Due to limitation in size, a sensor has only limited energy and storage. Traditional wireless network protocols cannot be applied in sensor networks directly. We study the distribution of events and queries in sensor networks. An event is something of interest detected by a sensor. A query is a request of information. A conventional approach to facilitate query nodes to acquire what they want is flooding. Nevertheless, flooding is not desirable in sensor networks due to the large number of nodes and limited energy in sensors. Recently, the concept of data-centric storage (DCS) is introduced where information of the same kind is kept in the same set of nodes. Queries can then be sent to these nodes for information retrievals. Theoretical analysis shows that this approach requires a lot fewer messages than the flooding approach when query frequencies are not high. Unfortunately, existing protocols based on DCS are prone to the hot-spot problem where some nodes have to handle lots of messages. In this paper, we present an efficient protocol for distributing events and queries in a location-aware sensor networks so that the load among nodes is more evenly distributed. We evaluate our protocol using simulations and the results show that our protocol successfully alleviates the hot-spot problem. © 2005 IEEE.