5

Consider a $M/E_2/1$ queueing system, where the customer arrival rate is $\lambda$ and the service time distribution has a gamma distribution with parameters $2$ and $\mu$, i.e. with p.d.f. $\mu^2te^{-\mu t}$ , $t ≥ 0$

(1) How can I determine the mean of the service time distribution?

(2) What is the traffic intensity $\rho$ of the system in terms of the parameters $\lambda$ and $\mu$?

My reasoning thus far: wouldn't I simply take the mean of a Gamma(2,$\mu$) distribution and thus just say the mean service time is $\frac{2}{\mu}$ but I'm guessing there has to be something more to it than that?

As for the triffic intensity $\rho$ I do not know what it would be for an erlang-2 queueing system? My thinking is (using $c=1$ for the number of servers):

$\rho = \frac{"mean.service.time"}{c*"mean.customer.interarrival.time"} = \frac{"mean.service.time"}{c*\frac{1}{"arrival.rate"}} =\frac{\frac{2}{\mu}}{1*\frac{1}{\lambda}}=\frac{2\lambda}{\mu}$

Note: Follow up question: Mean length of time spent queueing in $M/E_2/1$ system?

I am looking at question 5 for one of the 2012 RSS exams: http://www.rss.org.uk/uploadedfiles/userfiles/files/GD3_2012_final%20(web%20version).pdf

Community
  • 1
Clair Crossupton
  • 689
  • 7
  • 13
  • 1
    I don't think there's anything more to working out the mean service time than what you did. – Glen_b Jan 26 '14 at 02:59
  • @Glen_b Maybe I was over complicating it in my head. However, I don't know what the traffic intensity $\rho$ would be? – Clair Crossupton Jan 26 '14 at 10:56
  • Sorry, it's more decades than I'd like to mention publicly since I did any queueing theory -- so I don't remember what the definition of traffic intensity is, and google wasn't much help. If you could define it, perhaps, or point to one, that might help. (Incidentally, questions such as these would normally carry the `self-study` tag ([*q.v.*](http://stats.stackexchange.com/tags/self-study/info)); do you need all five of those tags? I am thinking you could manage without one of the `-process` tags) – Glen_b Jan 26 '14 at 22:04

1 Answers1

3

1) Yes, the mean of the service time distribution is just the mean of the Gamma(2,$\lambda)$ distribution.

2) The traffic intensity of the system is the arrival rate / the service rate, in this case:

$$\rho = \lambda \mu / 2$$.

These questions are a bit simplistic, but perhaps the intent of the first is to get you away from thinking of the mean service time as denoted by $\mu$ in all cases.

Interestingly enough, the Gamma distribution with integer shape parameter is also known as the "Erlang" distribution (hence the $E_2$ in the descriptor of the queuing system), and in telecommunications the various measures of traffic intensity (which are different than the queuing theory definition) are in "erlangs".

jbowman
  • 31,550
  • 8
  • 54
  • 107