Suppressing redundancy in wireless sensor network traffic

Abstract

Redundancy suppression is a network traffic compression technique that, by caching recurring transmission contents at receiving nodes, avoids repeatedly sending duplicate data. Existing implementations require abundant memory both to analyze recent traffic for redundancy and to maintain the cache. Wireless sensor nodes at the same time cannot provide such resources due to hardware constraints. The diversity of protocols and traffic patterns in sensor networks furthermore makes the frequencies and proportions of redundancy in traffic unpredictable. The common practice of narrowing down search parameters based on characteristics of representative packet traces when dissecting data for redundancy thus becomes inappropriate. Such difficulties made us devise a novel protocol that conducts a probabilistic traffic analysis to identify and cache only the subset of redundant transfers that yields most traffic savings. We verified this approach to perform close enough to a solution built on exhaustive analysis and unconstrained caching to be practicable. © 2010 Springer-Verlag.