ASACA: an adaptive service admission control algorithm to guarantee delay constraints in integrated services packet networks

We present an adaptive service (or call) admission control algorithm (ASACA) for integrated services packet networks. The proposed algorithm does not require a priori knowledge of the source characteristics, and it consists of two steps. The first step adapts the service rate of the system based on the number of frames lost during the observation period, and the second step estimates the expected delay associated with each class of service upon admitting or rejecting a service in the system. For this purpose, we use an M/M/1 queueing model such that the current statistical delay is added to the error estimates of the delay scaled by its gradient. While admitting a service, we ensure that the delay commitments towards the requirement profile of ongoing services are preserved. Further the sum of the statistical arrival rates for each class of service and the peak rate of the new call is not allowed to exceed the current system service rate. Simulation experiments demonstrate that our adaptive service control algorithm, ASACA, performs twice as well as a non-adaptive algorithm using an M/M/1 model, in meeting the service requirements but at the cost of a reduced number of calls admitted into the system, the number of which depends on the system load