Matching data dissemination algorithms to application requirements

A distinguishing characteristic of wireless sensor networks is the opportunity to exploit characteristics of the application at lower layers. This approach is encouraged by device resource constraints, and acceptable because devices are inexpensive and numerous enough that they can be dedicated to specific applications. Many data dissemination protocols have been proposed for multi-hop communication in sensor networks, each evaluated in some scenario. The premise of this paper is that, if protocols are designed to exploit application requirements, then no one protocol can be optimized for all applications.Instead, a family of protocols are needed, with guidance to match protocol to application. We show through field experiments with two tracking applications that choice of diffusion algorithm can affect application performance by 40--60%. These applications motivate the design of two new diffusion algorithms: push and one-phase pull diffusion. We describe these algorithms in comparison to previous algorithms, then systematically explore their performance as the number of sinks and sources, the traffic rate and node placement varies, and with and without geographic proximity in node placement and with and without geographically scoped communication. We characterize algorithm performance and highlight the effect of the choice of algorithm parameters. The end result of this work are guidelines to help application developers to match dissemination algorithms to application performance requirements.