Queueing analysis for input regulation method employing periodic monitoring and control (Q1068694)

Congestion control schemes are provided in telecommunications systems to prevent overload. This paper studies a queueing model for one of the control schemes and an input regulation method employing hysteresis control. In the method, excess traffic is regulated at the entrance to the network to keep the network load below capacity in order to avoid throughput degradation. Many studies have been made on the queueing models with hysteresis control employing continuous time monitoring. However, in real systems, input control activation is usually determined by the information obtained by monitoring such factors as resource utilization and the number of lost calls, at regular intervals. Input regulation models employing monitoring at regular intervals received little attention. The present paper assumes that the monitoring interval is arbitrarily distributed. Steady state probabilities at monitoring time points and those at arbitrary time points are obtained by using the piecewise Markov process theory. Some performance measures are calculated from these probabilities and numerical examples are given for the cases of regular interval monitoring, exponentially distributed interval monitoring and continuous time monitoring.