Generic selectors
Exact matches only
Search in title
Search in content

Reply To: Erlang Distribution


Am I correct in assuming you are wanting to simulate calls on/through the Base Station?
Erlang assumed that calls arrived in a uniform random manner, so the “Erlang Distribution” is really a uniform random distribution.
One way to simulate this is to divide the hour up into 1 second time slots and calculate the number of calls arriving in that second using a random number generator and the call per hour rate converted to calls per second by dividing by 3600, for example. Most computer software has random number functions that gives uniform random numbers.
You then have to use the random numbers the function generates to determine if there is a call made in the one second time slot by working out the probability of a call, or a number of calls in that one second period.
For example if the calling rate is 60 calls per hour, the probability of the call arriving in any particular second is 60/3600 or 1/60. If a call is generated, the probability of a second call at the same time is 1/60 x 1/60, and so on for the 3rd and subsequent calls while calls keep being generated in that second. In a particular simulation, more or less than 60 calls will be simulated, but the average will be 60 calls over repeated simulations.

Then once each call is set up, the call holding time will determine the probability that a call will finish in the next 1 second period. For example an average call holding time of 3 minutes or 180 seconds means the probability of the call finishing is 1/180 in any particular second. Statistically this means some calls may never hang up!

Erlang’s Equations allow all that simulation to be completely bypassed and calculates the dimensions of trunks directly from the statistical parameters.

Hope this helps.