Analysis on conformance deterioration for QoS provisioning: A stochastic approach

Abstract

Consider a scenario for QoS provisioning where two token buckets with identical token rate and bucket size are employed as the traffic shaper and policer respectively at the ingress and the egress nodes. Conformance deterioration is a non-intuitive phenomenon referring to the fact that an originally comformant packet may become non-comformant at the egress. We have earlier studied the phenomenon based on the assumption that the token status is static in [4]. In this paper, we continue to study a more general case where the token status is dynamic and propose a stochastic approach to analyze the probability of conformance deterioration, considering that the number of available tokens in the bucket is a random variable. The method is based on an equivalent queueing model for token bucket algorithm and is presented in a step-by-step manner. By defining four processes in the system model, we first calculate the probability that there are k tokens in the bucket and then obtain the probability of noncomformance at the ingress. The probability of conformance deterioration is further obtained by extending the analysis to the egress node, considering the influence of shaping delay and variable transmission time to the packet flow. Later in the paper, a concrete example is given to illustrate the proposed algorithm numerically. The study in this paper provides us with a useful means for analyzing the performance of traffic flows in a QoS capable network.