Abstract
This paper applies a matrix-analytic approach to analyze both the long-term and the short-term loss behaviors of a queue with space priority scheme. Five related performance measures are derived from conditional statistics, including the long-term high-priority and low-priority packet loss probabilities, and the three short-term measures - the average length of a critical period, the average length of a non-critical period, and the conditional high-priority packet loss probability during a critical period. The overall complexity of computing these performance measures is of the order O (K 2 m 1 3 m 2 3 σ 3 ), where m 1 , m 2 are the numbers of phases of the underlying Markovian structures for the high-priority and the low-priority packet arrival processes respectively and σ is the number of phases of the phase-type server. © 2006 Elsevier Ltd. All rights reserved.