G 0 0 taking into account the rates at which jumps

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: n +1) = nµ so ( /µ)n ⇡ (n) = ⇡ (0) n! If we take ⇡ (0) = e /µ. /µ and q (n, n 1) = then this becomes the Poisson distriibution with mean 4.4. EXIT DISTRIBUTIONS AND HITTING TIMES 133 Example 4.17. M/M/s queue with balking. A bank has s tellers that serve customers who need an exponential amount of service with rate µ nd queue in a single line if all of the servers are busy. Customers arrive at times of a Poisson process with rate but only join the queue with probability an if there are n customers in line. Thus, the birth rate n = an for n 0, while the death rate is ( nµ 0 n s µn = sµ n s for n 1. It is reasonable to assume that if the line is long the probability the customer will join the queue is small. The next result shows that this is always enough to prevent the queue length from growing out of control. Theorem 4.6. If an ! 0 as n ! 1, then there is a stationary distribution. Proof. It follows from (4.17) that if n ⇡ (n + 1) = If N is large enough and n P n · ⇡ (n) = an · sµ · ⇡ (n) N , then an /(sµ) 1/2 and it follows that ⇡ (n + 1) This implies that n µn+1 s, then 1 ⇡ (n) . . . 2 ✓ ◆n...
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