Regulation of a Connection Admission Control Algorithm

Connection Admission Control (CAC) algorithms are used to decide whether an incomingg connection shou ld be accepted or rejected in aa node of a network offering reservation based services in order to maintain the guaranteed Qualit y of Service (QoS) in the network. In this paper, we co nsider the statistical CAC algorithm proposed by Elwalid et al. [ 2 ]. The traffic model is made of ON-OFF sources and the QoS parameter is the loss probability. Based on the traffi c descriptors of existing and incoming connections, the algorithm takes its decision by computing an upper bound of thi s probability and checking whether it is larger than a gi ven tolerance ε. Usually this tolerance is aa fixed, given parameter. We propose here to adapt ε to react to the actual losses experienced at the node using a simple regulation mechanism: if th e actual loss rate is much smaller than the targeted loss rate, ε is increased to make a more aggressive usage of the available resources, and vice versa if the actual loss rate is t oo high. We discuss the influence of the regulation parameters and w e show that despite its simplicitythisregulatedCAC improve s significantly the performance of its non-tunable counte rpart. I INTRODUCTION We consider a networkk offering reservation-basedd servi ces. Connectionn Admissionn Control (CAC) algorithms are usedd to decide whether ann incoming connectionn shouldd be accepted or rejectedd inn a networkk node inn order to maintain the guaranteedd Quality of Service (QoS). Inn this paper, we consider the statistical CAC algorithm proposedd by Elwalidd et al. [ 2 ] andd describedd inn Section II. The traffic model is made of periodic ON-OFF sources and the QoS parameter i s the loss probability. Basedd on the traffic descriptor s of existing andd incomingg connections, the algorithm takes its decision by computing an upper boundd of this probability and comparingg it with a givenn tolerance ε. Usually this tolerance is a fixed, givenn parameter. In Section III we propose here to adapt ε to react to the actual losses experiencedd at the node using a simple regulation mechanism: if the actual loss rate is muchh smaller thann t he targetedd loss rate, ε is increasedd to make a more aggressive usage of the available resources, and vice versa if th e actual loss rate is too high. We discuss the influence of the regulationn …