Abstract
Performance evaluation is required at every stage in the life cycle of a network protocol. Analytical modeling is the method of choice for fast and cost effective evaluation of a network protocol. One of the most efficient high level formalisms used for analytical modeling of network protocols is Stochastic Petri Nets (SPN's). Yet, complexity of nowadays network protocols, which results in the state space explosion of the underlying Markov Chains (MC) model, has hindered the wide use of SPN's in the analysis of these protocols. Decomposition techniques are considered one of the main methodologies used for the approximate solution of the state explosion problem. Unfortunately, most of these techniques either require special form of the model or give insufficient accuracy of the solution. Moreover, small rates of communication protocols have caused additional overhead on any proposed technique for approximate evaluation of these protocols. This paper presents a novel decomposition technique called Behavior and Delay Equivalent Block (BDEB) technique. This technique overcomes most of drawbacks of other techniques proposed in the literature. It introduces a new aggregation method that depends on using a new type of transitions called intelligent transition. The proposed technique adopts delay equivalent criteria but in a new philosophy different from other techniques using the same criteria. The new technique is explained by applying it to a scalable model of the Courier protocol.