Stochastic

6 queueing networks canceling out the highest powers

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

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ure process are independent. Since there is more than a little hand-waving going on in the proof of Theorem 4.11 and its application here, it is comforting to note that one can simply verify from the deﬁnitions that Lemma 4.12. If ⇡ (m, n) = c m+n /(µm µn ), where c = (1 /µ1 )(1 /µ2 ) 12 is a constant chosen to make the probabilities sum to 1, then ⇡ is a stationary distribution. Proof. The ﬁrst step in checking ⇡ Q = 0 is to compute the rate matrix Q. To do this it is useful to draw a picture which assumes m, n > 0 (m 1, n + 1) (m, n + 1) I @ @ (a) @ @ µ1 @ µ2 @ @ (m 1, n) @ ? - (m, n) @ - (m + 1, n) I @ µ2 ? (m, n (c) @ 1) @ µ1 @ @ @ (b) (m + 1, n 1) The rate arrows plus the ordinary lines on the picture, make three triangles. We will now check that the ﬂows out of and into (m, n) in each triangle balance. In symbols we note that (a) µ1 ⇡ (m, n) = c m+n = ⇡ (m µm 1 µn 1 2 (b) µ2 ⇡ (m, n) = c m+n = µ1 ⇡ (m + 1, n µm µn 1 12 (c) ⇡ (m, n) = c m+n+1 µm µn 12 1, n) 1) = µ2 ⇡ (m, n + 1) This shows that ⇡ Q = 0 when m, n > 0. There are three other cases to conside...
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