Stochastic

nk k y rj nj j j 1 1 rj j 432 this

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: lity that both servers will stay busy for all time and the queue lengths will tend to infinity. Not covered by (i) or (ii) is the situation in which server 1 can handle her worst case scenario but server 2 cannot cope with his: 1 + p2 µ2 < µ1 and 2 + p1 µ1 > µ2 In some situations in this case, queue 1 will be empty often enough to reduce the arrivals at station 2 so that server 2 can cope with his workload. As we will see, a concrete example of this phenomenon occurs when 1 = 1, µ1 = 4, p1 = 1/2 2 = 2, µ2 = 3.5, p2 = 1/4 To check that for these rates server 1 can handle the maximum arrival rate but server 2 cannot, we note that 1 · 3.5 = 1.875 < 4 = µ1 4 1 · 4 = 4 > 3.5 = µ2 2 + p1 µ1 = 2 + 2 1 + p2 µ2 = 1 + To derive general conditions that will allow us to determine when a twostation network is stable, let ri be the long-run average rate that customers arrive at station i. If there is a stationary distribution, then ri must also be the long run average rate at which customers leave station i or the queue would grow linearly in time. If we want the flow in and out of each of the stations to balance, then we need r1 = 1 + p2 r2 a...
View Full Document

This document was uploaded on 03/06/2014 for the course MATH 4740 at Cornell.

Ask a homework question - tutors are online