Efficient Over-the-air Remote Reprogramming of Wireless Sensor Networks

Abstract

Over-the-air reprogramming is an important aspect of managing large wireless sensor networks. However, reprogramming deployed sensor networks poses significant challenges due to the energy, processing power and memory limitation of sensor nodes. For improved energy efficiency, a reprogramming mechanism should use less transmission and flash writing overhead. Past research has proposed different mechanisms for reprogramming deployed sensor networks. However, all of these mechanisms produce large patches if software modifications involve changing program layouts and shifting global variables. In addition, existing mechanisms use large amounts of external flash and rewrite entire internal flash. In this thesis, we present a differential reprogramming mechanism called QDiff that mitigates the effects of program layout modifications and retains maximum similarity between old and new software using a clone detection mechanism. Moreover, QDiff organizes the global variables in a novel way that eliminates the effect of variable shifting. Our experiments show that QDiff requires near-zero external flash, and significantly lower internal flash rewriting and transmission overhead than leading existing differential reprogramming mechanisms.