TCP congestion control with MDP algorithm for IoT over heterogeneous network

The expansion of Internet of Thing (IoT) will affect available bandwidth for transmission of packets across networks. Heterogeneous networks include many types of the network topology which help to optimize the congestion control in TCP Protocols. This causes packets to be delayed or lost in the entire network. We propose a Markov Decision Process (MDP) to enhances the state of congestion avoidance. Thus the MIMD mechanism is more efficient at controlling the congestion of window sizes than the TCP-Illinois or TCP-Scalable. The new optimize TCP congestion control protocol is named TCP-Siam. The TCP Siam is applied based on the TCP-Illinois hybrid protocol and modified with MDP state and rewarded when congestion avoidance gets to next state. TCP-Siam adds the coefficient to maximize the congestion window (cWnd) and improve the performance when package is lost due to a lossy link while connected to a WMN. TCP-Siam achieves good throughput, RTT-fairness, and utilization in a heterogeneous networks.