Scalable Redundancy for Sensors-to-Sink Communication

Abstract

In this paper, we present a new technique that uses deterministic binary network coding in a distributed manner to enhance the resiliency of sensor-to-base information flow against packet loss. First, we show how to use network coding to tolerate a single packet loss by combining the data units from k sensor nodes to produce k+1 combinations such that any k of them are solvable. After that, we extend the solution to tolerate multiple losses. Moreover, we study the coding efficiency issue and introduce the idea of relative indexing to reduce the coding coefficients overhead. To tolerate node or link failures, we introduce a simple routing protocol that can find maximally disjoint paths from the k sensor nodes to the base station (BS). We study the relationship between the probability of successful recovery of all data units at the BS, and the number of sources protected together taking into consideration their hop distance from the BS. From this study, we can decide on the appropriate number of sources to be protected together, so that the probability of successful recovery is higher than a certain threshold. Finally, we show through a simulation study that our approach is highly scalable and performs better as the network size increases.