INDIGO: An In Situ Distributed Gossip Framework for Sensor Networks

With the onset of Cyber-Physical Systems (CPS), distributed algorithms on Wireless Sensor Networks (WSNs) have been receiving renewed attention. The distributed consensus problem is a well studied problem having a myriad of applications which can be accomplished using asynchronous distributed gossip algorithms on Wireless Sensor Networks (WSNs). However, a practical realization of gossip algorithms for WSNs is found lacking in the current state of the art. In this paper, we propose the design, development, and analysis of a novel in situ distributed gossip framework called INDIGO. A key aspect of INDIGO is its ability to perform on a generic system platform as well as on a hardware oriented testbed platform in a seamless manner allowing easy portability of existing algorithms. We evaluate the performance of INDIGO with respect to the distributed consensus problem as well as the distributed optimization problem. We also present a data driven analysis of the effect certain operating parameters like sleep time and wait time have on the performance of the framework and empirically attempt to determine a sweet spot . The results obtained from various experiments on INDIGO validate its efficacy, reliability, and robustness and demonstrate its utility as a framework for the evaluation and implementation of asynchronous distributed algorithms.