A Low Collision Phase Synchronization Mechanism for Wireless Sensor Networks

Sensor network applications require nodes adjusted to the same phase to do communication scheduling, and to coordinate duty cycles. Inspired by the fireflies, which emit light in perfect synchrony, Mirollo and Strogatz first gave a theoretical model to model the firefly synchronicity phenomenon and prove its synchronization under two assumptions: 1. All-to-all topology and 2. No message delay or loss. Both of them make the theory impractical to real world applications i.e. all-to-all connection is not easily available and packets will lose due to congestion).
Geoffrey et al proposed the RFA (Reachback Firefly Algorithm) to compensate for the message delay problem. In the RFA, when nodes are under synchronicity, they will fire synchronizaing packets at the same time, and as a result, causes the synchronizing packets to collide. In this thesis, we propose a randomized algorithm, referred to as RandFA, to eliminate the collision problem.
RandFA and the state of art algorithms are implemented on the MSP430 platform for comparision. Our results show that RandFA experiences virtually zero packet loss, and the sensor node phases converge significantly faster to synchrony.