SDATP: An SDN-Based Traffic-Adaptive and Service-Oriented Transmission Protocol

Abstract

In this paper, a software-defined networking (SDN) based adaptive transmission protocol (SDATP) is proposed to support applications in fifth generation (5G) communication networks. For time-critical services, such as machine-type communication services for industrial automation, high reliability and low latency are required. To guarantee the strict service requirements, a slice-level customized protocol is developed with in-network intelligence, including in-path caching-based retransmission and in-network congestion control. To further reduce the delay for end-to-end (E2E) service delivery, we jointly optimize the placement of caching functions and packet caching probability, which reduces E2E delay by minimizing retransmission hops. Since the joint optimization problem is NP-hard, we transform the original problem to a simplified form and propose a low-complexity heuristic algorithm to solve the simplified problem. Numerical results are presented to validate the proposed probabilistic caching algorithm, including its adaptiveness to network dynamics and its effectiveness in reducing retransmission hops. Simulation results demonstrate the advantages of the proposed SDATP over the conventional transport layer protocol with respect to E2E packet delay.