Partially Observable Markov Decision Process for Transmitter Power Control in Wireless Sensor Networks

An agent based technique for controlling the transmission times and power of a wireless sensor network node is presented. The technique optimises the agent´s action in a non deterministic environment as is the case of a network node transmitting in the presence of interference from nearby nodes and obstacles in the environment. The formulated analysis is treated as discrete time and guarantees optimal results in a stationary partially observable environment.