Coordination of Large Distributed Sensor Networks Using Novel Evolutionary Algorithms

摘要

Real world systems such as advanced energy systems and smart grids are becoming more complex and interconnected. A critical issue in such systems is to ensure that the interactions of a large number of components do not negatively affect overall system performance. As these systems grow in complexity, modeling system components and their interactions becomes more challenging. Without accurate system models, system-wide sensing becomes critical, in order to accurately observe interactions between subsystems in order to monitor system performance. Distributed Sensor Networks (DSNs) offer a solution for system monitoring; however, as the number of sensors grows into the thousands, traditional control methods for these sensors which rely on hand-tuning or expert system designers become inadequate. In this work, we coordinate large sensor networks (up to 10,000 sensors) via evolutionary algorithms. This approach automates the development of controllers for the DSN, eliminating the need for hand tuning or human experts. We demonstrate the effectiveness, scalability, and reconfigurability of this approach for DSNs, and results show that our approach results in a 5 orders of magnitude increase in sensing performance over randomly generated sleep/sense schedules.