System level design issues in low-power wireless sensor networks

Abstract

Wireless Sensor Networks (WSN) are an emerging technology that is considered to have a high potential for realizing the vision of ambient intelligence. Tiny WSN nodes are unobtrusively embedded to environment for performing sensing, data processing, and actuating tasks. Capabilities of a single node are limited, but the feasibility of WSNs lies on the collaboration of nodes. WSNs are envisioned for a wide variety of applications ranging from home automation to military surveillance. Supporting the diversity of applications within the resource constraints is commonly considered to necessitate application-specific tailoring of nodes, communication protocols, and application algorithms.

This Thesis presents a design methodology for facilitating the development of application-specific WSNs from an abstract design phase to the prototype implementation. The functionality of a WSN is first designed in Wireless Sensor Network Simulator (WISENES) design environment with abstract deployment models. The novel features in WISENES are the graphical design of the models combined with the full-scale design time simulations for accurate evaluation of WSN performance, and the back-annotation of the performance results from prototypes for further improving simulator accuracy.

A runtime environment in the methodology preserves the design time abstractions during a prototype implementation on node platforms. The runtime environment for resource limited nodes is realized by a preemptive multithreading Operating System (OS), SensorOS, and a middleware that supports runtime distribution of application processing. A distinctive feature in SensorOS is its accurate time concept that enables the implementation of tightly synchronized WSN protocols and time sensitive applications. A lightweight allocation algorithm of WSN node middleware assigns application tasks and network maintenance roles with an objective to maximize network lifetime.

The application-specific tailoring of WSNs and the feasibility of the presented design methodology are illustrated by two case studies. An energy efficient Tampere University of Technology WSN (TUTWSN) targeted to low data rate monitoring applications is adapted to relay TCP/IP data through WSN and configured for delay sensitive operation in an indoor surveillance application. The results show that presented methods and tools facilitate and hasten the design, configuration, and implementation of WSN protocols and algorithms for different applications.