Notos: Efficient Emulation of Wireless Sensor Networks with Binary-to-Source Translation

Abstract

Developing for wireless sensor networks is a challenging task due to the severe resource constraints of the devices, the uncertainties of the environment, and the distributed nature of the system. Therefore, simulation is an essential tool for developing systems and for evaluating and comparing protocols at scale in a reproducible manner. Cycle-accurate emulation of sensor networks allows the execution of platform target code and provides deep insight into the behavior of the overall system including the important aspect of energy consumption. However, the required fidelity incurs a significant overhead and limits the size of the emulated networks considerably. We investigate the use of binary-tosource translation, where the machine code of an executable for the target platform is transformed to source code for the host platform and compiled as part of the emulator. Additionally, as part of this transformation we perform static analysis and optimize the generated code. We have implemented our approach based on the well-established emulator Avrora and show in our evaluation that this approach can lead to significantly higher simulation speeds.