Product-Form Approximate Solution for No-Product-Form Solution Stochastic Petri Nets

Dao-Bin LIU,Chuang LIN,Wei-ming LU

DOI: https://doi.org/10.3321/j.issn:0254-4164.2001.06.005

2001-01-01

Jisuanji Xuebao/Chinese Journal of Computers

Abstract:Stochastic Petri Nets(SPNs) are widely used for analyzing of complex systems such as computer systems, communication networks, and flexible manufacturing systems. Exact analytic solutions exist for a class of SPNs known as product-form solution SPNs. For this class of SPNs, the steady-state joint probabilities have a product-form solution. The performance parameters of product-form solution SPNs can be obtained using efficient algorithms such as the convolution algorithm and the MVA algorithm. Unfortunately, the assumptions of product-form solution SPNs are often too restrictive for modeling and analysis of real systems. Therefore, much work has been devoted to the approximate analysis of no-product-form solution SPNs. Most of the approximate methods are based on decomposition. This paper discusses the approximate solution of no-product-form solution SPNs.One possible approximate method is product-form approximation solution for no-product-form solution SPNs. The product-form approximation techniques are successfully used in Queueing Networks(QNs), such as Marie's method. The idea of Marie's method is to replace a subsystem by an equivalent exponential service station with load-dependent service rates. In this case, the parameters of the equivalent server are obtained by analyzing the subsystem in isolation under a load-dependent arrival process. The interarrival time is assumed to be exponentially distributed with rate λ(n), where n is the total number of customers currently presented in the network. The load-dependent service rates are then set equal to the so-called conditional throughput of the subsystem in isolation. Since the parameters λ(n) are unknown, an iterative procedure must be used to calculate them. Marie mainly applied his technique to the case of closed queueing networks with general service times. In that case, each subsystem consists of a single station which is analyzed in isolation as a λ(n)/GI/1/N queue. In this paper we introduce Marie's method to SPNs. The contributions of this paper are the following. We first extend the decomposition principles in Marie's method to more general cases using the results in SPNs and thus this technique can be applied to more cases. We formally describe these principles using operational analysis. Based on the formal description we provide mathematical proofs of the results in Marie's method. Finally, we give a justification of this method's accuracy by analyzing the sources of errors.