UMADE

From Cyber-Physical Systems Laboratory
Jump to navigationJump to search

Abstract

Wireless sensor networks have traditionally been used as specialized single-application platforms. However, with the emergence of integrated sensing systems such as automatic building management, wireless sensor networks are evolving into shared, multi-application infrastructure. This paradigm shift raises the need for a framework for deploying multiple applications on a wireless sensor network such that application QoS requirements are met while also optimizing the use of the limited sensor network resources. While several deployment and management toolkits have been developed for wireless sensor networks, none of these toolkits have the capability of addressing this problem. Hence, we propose a novel Utility-based Multi-Application Deployment Environment (UMADE) that supports the end-to-end process of (a) application QoS specification, (b) dynamic sensor node memory allocation that maximizes overall system utility, (c) automatic application deployment and (d) dynamic resource adaptation to network and application dynamics. We have implemented and evaluated UMADE on a wireless sensor network testbed of 28 motes, in the context of real applications required for building automation. Our evaluation results show that UMADE (a) successfully deploys multiple, dynamically arriving applications in a shared wireless sensor network, (b) achieves near-optimal sensor node allocations during each deployment, and (c) consistently improves the overall system utility. In addition to this we also present simulation results that evaluate the scalability of our system. Both our simulation and experimental results validate the feasibility and efficacy of utility-based multi-application deployment in shared wireless sensor networks.

Publications

People

This work is being conducted by Sangeeta Bhattacharya, a current PhD student, under the guidance of Dr. Chenyang Lu and Dr. Gruia-Catalin Roman.