Telephony system - calculating delayed message stats

36 Views Asked by At

I'm having a bit of an issue with a question likely to be on an exam I'm taking soon. Details are-

A messaging switching centre has three servers for passing messages to the network. If the servers are busy, the message passes into a queue for service on a first in first out basis. Message arrivals are Poissonian and server sesrvice times have a negative exponential distribution. During the busiest period, 400 messages arrive per minute. The average time taken to service a message is 5ms. Calculate the following:

  1. The fraction of messages that experience a delay
  2. The average delay for messages that experience a delay (in ms)
  3. The fraction of messages that are delayed for longer than 8.33ms during the busy period.

My process for trying to answer the first question is to calculate the probabilities of x or more calls arriving in a second, based on the 400/60 = 6.66 messages/second arrival rate-

$P(0) = \mu^x e^{-\mu} / x! = {6.66^0*e^{-6.66}}/0! = 0.0013$

$P(1) = 0.085$

$P(2) = 0.0283$

$P(3) = 0.0628$

So ultimately-

$P(x >= 4) = 1 - P(0)-P(1)-P(2)-P(3) = 0.8991$

So the chance of 4 or more calls in a second is 89.91%, but I'm just mind-blanking on how to then calculate the chance of calls being placed in the queue based on the service time.

The answers provided to us are-

  1. 0.444
  2. 2.22
  3. 0.01

Usually I'm pretty good at working backwards, but can't figure this one out. It's the method of this that I'm stuck with. Any help would be greatly appreciated.