Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Coordinated coverage in sensor networks via reinforcement learning

Kotzabasakis Georgios

Full record

---

Summary

Machine Learning is a fast developing and ever growing field in computer science. In addition to that, Sensor Networks are also a very promising field that has significant impact on a variety of applications. Given these facts, a multi-agent system (MAS) approach on wireless sensor networks (WSNs) comprising sensor-actuator nodes is very promising, as it has the potential to tackle the resource constraints inherent in these networks by efficiently coordinating the activities among the nodes. Furthermore, a very common issue in the field of sensor networks is the sensing coverage problem, which is the task of properly and sufficiently covering an area. In this thesis, we consider the coordinated sensing coverage problem and study the behavior and performance of the fully distributed Q-Learning algorithm for reinforcement learning using linear value function approximation. We use the Tossim platform to simulate our TinyOS application, which consists of different topologies of sensor networks with parametric sizes. Subsequently, we present the results of our simulation and display a number of graphs to visualize performance and learning outcomes on three specific topologies. We consider issues, such as successful convergence to optimal policies and maximization of local and global rewards. The implementation results are quite promising, since our algorithms exhibit high percentage of successful convergence to optimal policies.

Available Files

Services