- This topic has 1 reply, 1 voice, and was last updated 20 years ago by R.
12th February 2001 at 14:40 #19959kawfeeGuest
Hello, I recently inquired about the Erlang B Model. I am interested in thoughts on using the Poission traffic model as compared to the erlang B model for Voice over IP.
I believe Erlang B to be a more appropriate model, but am still researching. Ultimately, which ever model is used, the “blocking” variable has to be set to a bare minimum, since voice traffic must always get 1st priority over data.
Any comments on modifying one of the models, so that blocking is virutally eliminated, and in it’s place a new variale is added, that will essentially use packet size (or packet duration) as a blocking variable substitute. Essentially, if a voice packet tries to go through, just as a data packet has been sent, the maximum time the voice packet will have to wait before it can be sent even given its priority status, is the duration of that already sent data packet (i.e. 10ms) Data packets size are typically kept small, just so that this scenario doesn’t cause a voice packet to have wait any substanial time to be sent, but in a sense, that voice packet will have to wait at maximum, the duration of any data packet already sent before it to pass. So, I was thinking of using that as a new variable instead of the traditional blocking variable, which in a VoIP network, isn’t typically desired (blocking of calls that is).
Questions, comments, stories?!?12th February 2001 at 23:37 #19960RGuest
It sounds like you are looking to employ the erlang C model in a data network. It can be used, but since the formulae are based on averages and assumptions you would have to base it on an average packet size.
Erlang C model defines the probability of delaying or queuing a call…in this case a packet.