MIT6_262S11_lec20

# Useful application all birthdeath processes with j pj

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: ible if and only if the embedded chain is. 7 The guessing theorem: Suppose a MP is irreducible and {pi} is a set of probabilities that satisﬁes piqij = ￿ pj qji for all i, j and satisﬁes i piνi < ∞. Then (1), pi > 0 for all i, (2), pi is the sample-path fraction of time in state i WP1, (3), the process is reversible, and (4), the embedded chain is p ositive recurrent. ￿ Useful application: All birth/death processes with j pj νj < ∞ are reversible. Similarly, if the Markov graph is a ￿ tree with j pj νj < ∞, the process is reversible. 8 ✉ ✉ a2 d1 a1 ✉ ✉ a3 d2 ✉ d3 ✉ ✉ a4 d4 ✉ Right moving (forward) M/M/1 pro cess ✉ d3 a4 d4 ✉ ✉ d2 ✉ ✉ a3 ✉ a2 d1 ✉ ✉ a1 Left moving (backward) M/M/1 process Burke’s thm: Given an M/M/1 queue in steady-state with (arrival rate) λ < µ (departure rate), (1) Departure process is Poisson with rate λ (2) State X (t) is independent of departures b efore t (3) For FCFS, a customer’s arrival time, given its de­ parture at t, is independent of departures b efore t. 9 ✉ ✉ a2 d1 a1 ✉ ✉ d3 a4 d4 ✉ ✉ ✉ a3 d2 ✉ d2 ✉ ✉ a3 t d3 ✉ t ✉ a2 ✉ a4 d1 ✉ d4 ✉ ✉ a1 A departure at t in (right-moving) sample path is an arrival in the M/M/1 (left-moving) sample path. For FCFS left-moving process, departure time of ar­ rival at t depends on arrivals (and their service req.) to the right of t; independent of arrivals to left. For corresponding (right-moving) process, the arrival time of that departure is independent of departures b efore t. 10 P...
View Full Document

## This note was uploaded on 01/13/2012 for the course ELECTRICAL 6.262 taught by Professor Staff during the Fall '11 term at MIT.

Ask a homework question - tutors are online