Dynamic Load Balancing in Sensor Networks

Abstract

A sensor network with a tree topology faces challenges in efficient data aggregation, especially in normalizing energy consumption. Our paper introduces a new load-balancing algorithm that works autonomously to reduce overload and to distribute energy requirements without an energy-aware module in the sensor. We investigate the differences produced by two approaches to decision-making: a semi-centralized approach or a fully-distributed one. We also discuss how to avoid cascading reorganization as loads are balanced.