Article ID: | iaor201526846 |
Volume: | 23 |
Issue: | 4 |
Start Page Number: | 477 |
End Page Number: | 496 |
Publication Date: | Jun 2015 |
Journal: | International Journal of Operational Research |
Authors: | Jain Madhu, Bhagat Amita |
Keywords: | markov processes, simulation |
In the present investigation, we study the maximum entropy principle (MEP) for the Mx/G/1 retrial queue with general retrial times and unreliable server. The customers arrive in bulk in Poisson fashion and are served if they find the server idle otherwise join the virtual pool of customers called orbit. By using maximum entropy principle, approximate results for the queue length as well as waiting time are obtained. We construct the maximum entropy function by using several known constraints so as to determine the probability distributions. Moreover, a comparative study has been done between the exact analytic expressions obtained by the supplementary variable technique and approximate results obtained by MEP.