Online Parameter Adaptation for Sensor Network MAC Protocols

Abstract

The performance of sensor network MAC protocols is heavily dictated by their operating parameters. To operate efficiently, these protocols must continuously adapt to wireless link and traffic dynamics. We present a framework to adapt the MAC parameters in an online fashion. Specifically, our framework allows users to specify application requirements as objectives and constraints on three network-wide performance metrics: network lifetime, end-to-end reliability, and end-to-end latency. Unlike previous work, we formulate the resulting parameter optimization problem as a multi-objective optimization problem. We use well-established techniques and tools for constraint programming to identify an optimized solution depending on the current network state. This includes a vector of MAC parameters installed on the sensor nodes at runtime. Testbed experiments with the X-MAC protocol demonstrate that our framework significantly improves the system performance under user-defined constraints as the network conditions evolve. In particular, our approach avoids overprovisioning the MAC configuration based on the highest expected traffic load—as is common practice in real-world deployments—achieving lifetime gains of more than 30% in some of our experiments.