시스템 성능 평가 퀴즈8

수 25 5월 2016

Q. A machine shop has four machines, A, B, C, and D. The numbers of servers in the machines A, B, C, and D are one, one, two, and three, respectively. Service time distributions of the servers in the machines A, B, C, and D are exponential at their respective rates \(\mu_a\), \(\mu_b\), \(\mu_c\), and \(\mu_d\). The shop gets four types of jobs, numbered 1 through 4, where each type requires service on machines in a particular sequence; type 1: ABDA, type 2: CABC, type 3: ACB, type 4: BCAD. The arrival process of type \(i\) jobs is Poisson at rate \(\lambda_i\).

(a) Under what conditions is this system stable?

(b) What is the joint stationary distribution of the number of jobs at each machine?

A.
주어진 시스템은 시스템으로 job이 들어오고, 일련의 과정을 거쳐서 job이 나가며, 여러 machine으로 구성된 Open Queueing Network이다. 또한 arrival rate가 Poisson process이고 모든 server가 exponential이므로 Product form을 생각할 수 있다. 한편 시스템이 안정되려면 service rate가 arrival rate보다 작아야 한다.

(a) 안정된 시스템이 되려면 service rate가 arrival rate보다 작아야 하므로, 임의의 machine S에 대해서 arrival rate를 \(\lambda_s\)라고 할 떄, 다음이 성립해야 한다.

$$ \lambda_a < \mu_a $$
$$ \lambda_b < \mu_b $$
$$ \lambda_c < 2\mu_c $$
$$ \lambda_d < 3\mu_d $$

(b)

$$ P(n_A, n_B, n_C, n_D) = P_A(n_A)P_B(n_B)P_C(n_C)P_D(n_D)$$

이때 각 서버는 다음과 같다.
A : \(M/M/1\), \(P_A(n_A) = (\frac{\lambda_a}{\mu_a})^{n_A}(1-\frac{\lambda_a}{\mu_a})\)
B : \(M/M/1\), \(P_B(n_B) = (\frac{\lambda_b}{\mu_b})^{n_B}(1-\frac{\lambda_b}{\mu_b})\)
C : \(M/M/2\), \(M/M/m\) 참조
D : \(M/M/3\), \(M/M/m\) 참조

한편 \(M/M/m\)에서 arrival rate와 departure rate는 같은 점을 이용하면 다음을 알 수 있다.

$$ \lambda_a = \lambda_1 + \lambda_1 + \lambda_2 + \lambda_3 + \lambda_4$$
$$ \lambda_b = \lambda_1 + \lambda_2 + \lambda_3 + \lambda_4$$
$$ \lambda_c = \lambda_2 + \lambda_2 + \lambda_3 + \lambda_4$$
$$ \lambda_d = \lambda_2 + \lambda_4$$



위의 것들을 조합해서 답을 내면 되는 것 같은데 잘 모르겠으니 패스

Category: quiz Tagged: system analysis quiz markov chain transition probability diagram