Intelligent End-to-End Congestion Control in Wireless Mesh Networks

Wireless mesh networks have unique characteristics such as the lossy nature of the communication medium, absence of a base station, etc and due to this, maintaining the performance of reliable transport protocols, such as TCP, over wireless mesh networks is a challenging problem. One of the reasons for poor performance of conventional TCP variants over WMN is that the congestion control mechanisms in conventional TCP variants do not explicitly account for these unique characteristics. To address this problem, a novel neural network based congestion control technique is proposed for reliable data transfer over WMN. The proposed technique will be analyzed in detail and is incorporated into TCP to create a variant that we name as intelligent TCP or iTCP. Also, in a wireless communication, any link failure will suspend data delivery completely and thus the overhead of route discovery cannot be neglected. The routing should take place reducing the routing overhead providing a suitable rebroadcast delay for determining the rebroadcast order and thus decreasing the number of retransmissions. So the objective is to control the congestion while providing a more flexible routing to improve the routing performance, while not to violate the basic principles of proposed protocol. The performance of iTCP is evaluated using ns-2 simulations and the results demonstrate that our proposed technique exhibit significant improvement in total network throughput and average residual energy. Keywords— Wireless mesh networks, transmission control protocol, neural networks, congestion control, upstream