Adaptation Service Framework for Wireless Sensor Networks with Balanced Energy Aggregation

Abstract

Wireless sensor networks consist of tiny, energy-constrained sensor nodes that may be deployed in large numbers. Considering these unique characteristics, our adaptation service framework is designed to deliver distributed events and react to changes through distributed actions. This adaptation service framework makes use of energy efficient data aggregation which helps to maximize network lifetime under limited energy constraints. The architecture for efficient adaptation service with distributed events consists of three type of components: event sensor, adaptation server, and action node. An event sensor is responsible for detecting, collecting, and sending events to an adaptation server. An adaptation server receives events from various event sensors and sends action requests to action nodes. An action node executes the requested action and replies to the adaptation server when it has completed the action. Our research primarily focuses on energy-efficiently delivering the large amount of events by building on prior work done on data aggregation over Directed Diffusion. Although using aggregation paths is energy efficient, nodes at aggregation points consume more energy than any other nodes since they are expected to have greater load and processing overhead for aggregating events. As a result, these nodes have a shorter system lifetime than others. Our new aggregation scheme, called balanced energy aggregation, focuses on maximizing the overall lifetime of the sensor network. When a node at an aggregation point is overloaded, the next closest node to the sources on a shared path is selected as a new aggregation point. This allows for energy to be saved by distributing loads across different paths from the sources to the sinks. Our programming model is based on a concept of channels which encapsulate properties of the underlying communication system, i.e. a data forwarding path with aggregation points for conserving energy and maximizing lifetime. Channels are accessible to applications for communication used between adaptation servers and action nodes as well as event sensors and adaptation servers. The algorithms have been successfully implemented and tested over Directed Diffusion protocol. Our experimental results demonstrate that the proposed aggregation algorithm outperform previous methods such as Greedy Aggregation and Path Sharing Aggregation in terms of system lifetime, average dissipated energy and number of events.