Abstract

In this paper, the prototype implementation of a scalable, distributed protocol for calculating the global average of sensed environmental variables in unattended wireless sensor networks (WSNs) is presented. The design and implementation of the protocol introduces a communication scheme for discovering the WSN topology. Such scheme uses a synchronous flooding algorithm, which was implemented over an unreliable radiogram-based wireless channel. The topology discovery protocol has been synchronized with sampling time of the WSN and must be executed before the consensus-based estimation of the global averages. An average consensus algorithm, suited for clustered WSNs with static topologies, was selected from the literature. The algorithm was properly modified so that its implementation guarantees that the convergence time is bounded and less than the sampling time of the WSN. Moreover, to implement the consensus algorithm, a reliable packet-passing protocol was designed to exchange the weighting factors among the sensor nodes. Since the amount of data exchanged in each packet is bounded by the degree of the WSN, the scalability of the protocol is guaranteed to be linear. The proposed protocol was implemented in the Sun SPOT hardware/software platform using the Java programming language. All the radio communications were implemented over the IEEE 802.15.4 standard and the sensed environmental variables corresponded to the temperature and luminosity.