Rapid Development and Flexible Deployment of Adaptive Wireless Sensor Network Applications

  • Authors:
  • Chien-Liang Fok;Gruia-Catalin Roman;Chenyang Lu

  • Affiliations:
  • Washington University in St. Louis;Washington University in St. Louis;Washington University in St. Louis

  • Venue:
  • ICDCS '05 Proceedings of the 25th IEEE International Conference on Distributed Computing Systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

Wireless sensor networks (WSNs) are difficult to program and usually run statically-installed software limiting its flexibility. To address this, we developed Agilla, a new middleware that increases network flexibility while simplifying application development. An Agilla network is deployed with no pre-installed application. Instead, users inject mobile agents that spread across nodes performing application-specific tasks. Each agent is autonomous, allowing multiple applications to share a network. Programming is simplified by allowing programmers to create agents using a high-level language. Linda-like tuple spaces are used for inter-agent communication and context discovery. This preserves each agentýs autonomy while providing a rich infrastructure for building complex applications, and marks the first time mobile agents and tuple spaces are used in a unified framework for WSNs. Our efforts resulted in an implementation for MICA2 motes and the development of several applications. The implementation consumes a mere 41.6KB of code and 3.59KB of data memory. An agent can migrate 5 hops in less than 1.1 seconds with 92% reliability. In this paper, we present Agilla and provide a detailed evaluation of its implementation, an empirical study of its overhead, and a case study demonstrating its use.