A learning based protocol for bidirectional congestion control in wireless sensor network

One of the most important challenges in WSNs is how to address congestion problem in such environments. The nature of WSN applications such as data centric and many to one data flow and infrastructure of sensors offers different solutions. In some particular applications of WSN such as reprogramming nodes or sending commands to them it is necessary that the sink is able to send data to the sensors in the least possible time. So in such stats, in addition to solving congestion in upstream direction, downstream congestion control is also needed. Although there are approaches that work on congestion in sensor to sink traffic but there is no protocol that control bidirectional congestion. In this paper an approach based on learning automata is proposed. Each intermediate node has an automaton and makes decision to increase or decrease upstream and downstream data rate according to feedback signal which received from environment that sensor works on it.