TCP Performance Enhancement Using Cross-Layer Method in IoT Networks

Transmission Control Protocol also known as TCP, utilizes many paths for the simultaneous transmission of data in order to increase its overall performance. But, the previous TCP had a problem to transmit a larger number of subflows in the Internet of Things (IoT) networks. We used a cross-layer design to enhance TCP performance in the IoT networks by implementing effective packet scheduling & congestion control in order to address the existing issues. The IoT network uses grid topology and Manhattan distance to improve scalability and flexibility. After network configuration, packet scheduling is conducted using proportional fair scheduling with fitness mechanism (FBPF) and choosing best subflow to minimize transmission latency. Scheduled subflows are sent along optimal paths to increase goodput and throughput. After packet scheduling, the proposed algorithm is used to control congestion in TCP. This algorithm consists of three stages: congestion detection stage, fast retransmission, and fast recovery. To improve fast retransmission time, we utilize a cat and mouse based optimization algorithm (CMBO). The adoption of a metaheuristics algorithm improves the convergence speed thereby improving the user throughput. Simulation findings indicate that our proposed scheme outperforms high TCP performance with respect to goodput, throughput, packet loss, congestion window size, and transmission delay.