Revenue optimization and fairness control of priced guaranteed and best effort services on an ATM transmission link

This paper is the sequel of two previous papers describing and analyzing partially blocking-queueing systems. In those papers Blaabjerg et al. (see Proc. of the IEEE International Conference on Communication Systems, ICCS '96, Singapore, 1996 and Proc. of the 5th IFIP International Conference on Telecommunications Systems Performance Modelling, Nashville, TN, USA, 1997) proposed the extension of the classical multirate circuit switched loss model in order to model and analyze on the call level ATM systems supporting best effort type service classes, such as the ABR or UBR service classes. After a brief review of the basic modelling concepts, in this paper we are concerned with optimal call admission control, (CAC) of CBR/VBR and ABR/UBR calls in the sense that we want to maximize the revenue on a single ATM link when both QoS guaranteed and best effort service classes are present. We formulate the problem as a Markov decision problem and find that the concept of intelligent blocking is useful in this context. We solve this optimization task by policy iteration to find the blocking states of the link. This modelling framework (in its present form and by extensions) allows one to study the impact of pricing policies on CAC and fairness.