Abstract
In this paper, a reinforcement-learning-based scheduling framework isproposed and implemented to optimize the application-layer quality-of-service(QoS) of a practical wireless local area network (WLAN) suffering from unknowninterference. Particularly, application-layer tasks of file delivery anddelay-sensitive communication, e.g., screen projection, in a WLAN with enhanceddistributed channel access (EDCA) mechanism, are jointly scheduled by adjustingthe contention window sizes and application-layer throughput limitation, suchthat their QoS, including the throughput of file delivery and the round triptime of the delay-sensitive communication, can be optimized. Due to the unknowninterference and vendor-dependent implementation of the network interface card,the relation between the scheduling policy and the system QoS is unknown.Hence, a reinforcement learning method is proposed, in which a novel Q-networkis trained to map from the historical scheduling parameters and QoSobservations to the current scheduling action. It is demonstrated on a testbedthat the proposed framework can achieve a significantly better QoS than theconventional EDCA mechanism.