A Spatial Computing Approach to Programming Large Scale Wireless Sensor Networks

Abstract

The technology required to design and deploy large scale wireless sensor networks is available, affordable and shows an increased interest in various application domains. However, application development for distributed systems is a cumbersome task, typically carried out with low-level embedded programming paradigms. A middleware is usually employed to bridge the gap between the node-level programming and the overall-system application development. Sometimes, the later one is even performed by non-IT specialists. In this thesis we collaborate with the Chess company to integrate the spatial-computing-paradigm-based Proto middleware with the MyriaNed production-ready wireless sensor network. We analyze the integration challenges and provide a mechanism that integrates the Myria specic characteristics (e.g., Gossiping-based communication protocol, Myria Core software design) with Proto. Additionally, we implement new software modules (viral code dissemination, target-oriented adjustment of parameters and a wireless network snifer for debugging) to enhance the application development process. Finally, we create applications using Myria-Proto to check the capabilities of the new software stack.