A Fuzzy-Based Servers’ Management Technique for M/M/m Queue Network

Abstract

Abstract The study presents an M/M/m queue model for a fuzzy-based management of parallel multiple servers. In this network, customers arrive according to a homogeneous Poisson process at rate λ and are served on first-come, first-served basis. The model may have up to three categories of servers viz: active servers, ’server(s) on vacation’ and reserved servers. In a situation where all active servers are busy and customers arrive, there are three possibilities: a server on vacation (if any), is turned on to attend to the arriving customers; the system request for additional server(s) from among reserved servers or the arriving customer(s) wait until one or more of the active servers becomes idle. Decision on management of servers is based on the output. This decision is taken by an Expert System which adopts fuzzy logic approach to process inputs as crisp variables to generate output based on given rules. This is necessary to ensure that available servers are put to optimal use while arriving customers are serviced within the shortest possible time and servers’ costs are minimized.