Birth and Death Processes Example: Communication System • A communication system has two processers that decode messages and a buffer that holds at most two messages. • A message that arrives when the buffer is full is lost. • Each processer needs an average of 2 minutes to decode a message. • Messages arrive at a rate of 1 per minute. Assume that inter-arrival times and processing times are independent exponential random variables. (a) Draw a state diagram with possible states and corresponding birth/death rates. Solution: 1 1 1 0 .5 1 1 2 3 1 4 1 1 Let T1 = time until processer 1 finishes and T2 = time until processer 2 finishes. E[Ti ] = 2 i = 1, 2 implies that rate parameter for both distributions are 1/2 i.e., T1 , T2 ∼ iid Exp(1/2) For k = 1, implying that µ1 = 21 = .5 For k ≥ 2, D = min{T1 , T2 } giving D ∼ Exp( 12 + 1 2 = 1); thus µk = 1, (b) What is the (large t) probability that the system is empty? S =1+ implying that p0 = 1 S = 1 1·1 1·1·1 1·1·1·1 + + + =9 .5 1 · .5 1 · 1 · .5 1 · 1 · 1 · .5 1 9 (c) What is the pmf of X? pk = P [X = k] = k pk = P (X = k) 1 λ0 λ1 . . . λk−1 p0 µ1 µ2 . . . µk 0 1 2 3 4 1 9 2 9 2 9 2 9 2 9 k = 2, 3, 4 (d) What is the (large t) expected # of jobs in the system? E[X] = 0 · p0 + 1 · p1 + 2 · p2 + 3 · p3 + 4 · p4 1 2 = 0 · + · (1 + 2 + 3 + 4) 9 9 = 20/9 (d) What is the probability that an incoming in message is turned away? An incoming message is lost when system is full i.e., X = 4 P (message lost) = P (system full) = P (X = 4) = p4 = 2/9 (e) What is the expected number of messages in the buffer? Let Y = # of messages in the buffer; note that Y is a function of X X Y pk 0 0 1 0 2 0 3 1 4 2 1 9 2 9 2 9 2 9 2 9 E[Y ] = 0 · p0 + 1 · p1 + 2 · p2 + 3 · p3 + 4 · p4 2 2 2 2 1 = 0· +0· +0· +1· +2· 9 9 9 9 9 = 5/9 2