Wireless sensor network (WSN) is a growing technology used to perform different tasks. Due to their source limitation such as lack of memory, computation resource and power, WSNs are susceptible to faults and to be congestion which is one of the its main problems. Congestion causes Quality of Service (QoS) degradation due to delay, energy consumption and packets loss which decreases the network lifetime.
In this paper, an adaptive neuro-fuzzy inference system (ANFIS) congestion control protocol is proposed and tested on an IParking that we developed. The congestion is detected in the sink node and controlled through it by adjusting the sending rate in the source nodes. Local parameters were used to estimate congestion like participants, traffic rate and buffer occupancy. We used a real application to evaluate our protocol: parking service that is easy to implement in existing parking lots, low cost and provided with an efficient sensor to track vehicles. Our results prove the efficiency and the reliability of our protocol.