|
|
|
An extremely well written paper that discusses routing problems in wireless sensor networks. This is a new topic that is guaranteed to generate a lot of interest. A sensor network is a collection of sensors that capture, process and disseminate sensing information such as terrain images, seismic data, etc. The data on the sensors of the network constitute a distributed database. For example, in the case of sensors that are parachuted or sprayed from an airplane, the database renders a global picture of an unknown terrain from local images collected by individual sensors. The problem addressed in the paper is how replicate the database at each one of the sensors, i.e., how to send each piece of information obtained by a sensor to all the other sensors. A straightforward solution is called flooding: when receiving a new data item each node stores it in its local database, and also sends it to all its neighbors. The paper explains convincingly why this is inappropriate in sensor networks which are resource- (energy, computation, communication) poor. Then the paper proposes and analyzes protocols based on three types of messages: advertise a data item, request a data item, send a data item. These protocols seem new and interesting, but they seem to repeat a network discovery process from each sensor, for each new data item that the sensor obtains. An alternative straightforward scheme that should be considered is the following. First each node finds the topology of the network, and then it communicates each new data item to all the other sensors along some (minimum) spanning tree.
Overall the paper is interesting and provides a good introduction to a novel and promising area.
Copyright © 2000 by the author(s). Review published with permission.