Optimization of buffer overflow probability in Jackson queueing networks using Mamdani fuzzy inferecnce system

In this paper we consider Mamdani fuzzy inference system to study for optimizing the buffer overflow probability in a single M/M/1 queue and two M/M/1 queue in tandem Jackson networks. Mamdani fuzzy inference system is proposed to estimate the service rate for the Jackson queueing network. The arrival rate and the highest overflow level of a particular queue are provided to the Mamdani fuzzy inference system which generates the service rate according to the fuzzy rule base. The arrival rate, the highest overflow level and the new service rate will be used to estimate the buffer overflow in a single M/M/1queueing network then in two queues in tandem queueing network. Simulation results shows that Mamdani fuzzy inference system reduce the buffer overflow probability as compared with the normal estimation of buffer overflow probability.