Energy-aware software updates in heterogeneous Wireless Sensor Networks

Software updates for Wireless Sensor Networks have traditionally taken the form of either binary machine code or bytecode for a target Virtual Machine. Each aim to strike a compromise between radio transmission power and CPU execution time for the target microcontroller. Binary updates are either monolithic or modular, while bytecode updates are modular. We present a novel paradigm which takes advantage of both these approaches' strengths while alleviating their weaknesses. Our solution consists of compiling in-situ (on the WSN node) a high-level language called MinTax. Next, we demonstrate this approach in an heterogeneous context, consisting of WSN nodes of different hardware architectures. The sensor nodes we deployed our solution onto are MSP430 hardware-based(16-bit core) and AVR hardware-based(8-bit core).