A PiggyBacking Approach to Reduce Overhead in Sensor Network Gossiping

Many wireless sensor network protocols are employing gossip-based message dissemination, where nodes probabilistically forward messages, to reduce message overhead. The authors are concerned with emerging systems in stationary sensor networks that are multiple-source with each message targeted at every recipient, such as query and code propagation. Default gossip-based approaches tend to treat each stream of messages from different senders independently of the others, overloading each node with message overhead summed from all streams. They apply intelligent scheduling strategies for gossip forwarding, effectively piggybacking streams atop one another, to address this significant message overhead. The problem formulation introduces a new concept called the "Semblance graph" used to schedule gossiping based on streams' gossip periods.