GA-based PID active queue management control design for a class of TCP communication networks

Abstract

Active queue management (AQM) is a key congestion control scheme for reducing packet loss and improving network utilization in TCP/IP networks. This paper proposes a proportional–integral-derivative (PID) controller as an active queue manager for Internet routers. Due to the limitations of packet-dropping probability and the effects of propagation delays in TCP networks, the TCP AQM network was modeled as a time-delayed system with a saturated input. An improved genetic algorithm is employed to derive optimal or near optimal PID controller gains such that a performance index of integrated-absolute error (IAE) is minimized, and thereby a stable queue length, low packet loss, and high link utilization for TCP networks are guaranteed. The performance of the proposed control scheme is evaluated in various network scenarios via a series of numerical simulations.