ABSTRACT
Usage Parameter Control (UPC), one of the most fundamental preventive congestion control mechanisms, monitors the traffic generated by the sources and enforces the misbehaving ones to abide by the traffic contract negotiated during the call set-up phase. Most of the known UPC mechanisms do not succeed in detecting every possible violation of the traffic contract, while their reaction time is in some cases unacceptable for real-time applications. Consequently, they cannot enforce the source to behave compliantly, while the excess undetected traffic deteriorates the Quality of Service (QoS) requirements of the conforming users in an internodal node. In this chapter, after a brief review of the most known UPC mechanisms and a more detailed description of the Leaky Bucket, the best known of all, a UPC mechanism that uses a Reinforcement Learning Algorithm to enhance the performance of the Leaky Bucket, is proposed. A simulation study is used to show the effectiveness of this mechanism both in enforcing suitably and quickly the misbehaving sources and in guaranteeing the QoS parameters of the conforming sources. Finally, the hardware implementation concept is discussed and our conclusions are drawn.
