- This topic has 24 replies, 1 voice, and was last updated 15 years, 3 months ago by colein.
21st December 2001 at 18:54 #32389JuanGuest
I need the phone of Ricardo Arcadi16th February 2002 at 23:15 #32390basemGuest
i am in engineering faculty and i want to design atelecom25th March 2002 at 14:35 #32391stephenGuest
my problem is that i have switch capable of handling 6000 erlangs of data traffic. however in all earlier cases ive seen i can only quantify a nodes capacity using throughput in kbit/sec. so how would i change this figure in erlang into kbps24th August 2002 at 12:29 #32392davidGuest
I would like to know if you got a response to your question. I am also seeking an answer to similar question. Many thanks.29th August 2002 at 03:22 #32393xiao xiaoGuest
I agree Richard Arcadi, but add following:
M/M/1 is only for some applications, e.g. UDP traffic;
M/G/1-PS queuing model shoud be used for long-range dependency and self-similarity application, such as HTTP, Telnet, FTP.17th November 2002 at 07:11 #32394nasirGuest
may i have email Mr Rommel pls.25th November 2002 at 05:03 #32395Ali Hammad AkbarGuest
I am in NUST, Pakistan. Question: How can we possibly relate a traffic load of 2.8 Erlangs to bps or Packets/sec on ethernet [Packet Size = 1500 Bytes]? I mean that Erlang is basically a PSTN traffic unit. In order to map it to data networks, we need to know certain information as Analog-to-Digital Encodinh scheme, interpacket delay etc. Can somebody help? Mr Rommel, can you please provide a basis? Thanks Ali26th November 2002 at 01:15 #32396RommelGuest
In my example above, I was just stating a way it could be used. If you are tied down to the symantics of 1 Erlang meaning the total time it takes to keep a 64k channel busy for one hour than it would not work. If you redefine erlang to mean something like the amount of traffic it takes to keep a traffic sensitive facility busy for one second then it can work. If you look at the above example I laid out how it possibly could be used. In all reality since it only takes into account the randomness of traffic and not the peakedness of traffic then the calculations will be off.
We need something that will take into account peakedness as well as a blocked call disposition of queued.
Since I am a voice guy, and used to voice methods, I have not done any research into other methods of traffic analysis. I am sure there are other ways around it. But this is a voice guy’s view of it when going into a data world. In all reality, most folks do not do anything more than add bandwidth when their networks begin to slow.5th December 2002 at 08:38 #32397AnjumGuest
This is regarding using Erlang B for packet data networks.
I have cellular packet data network in which packets originating from user follow a dedicated path via paricular card on BTS , then on particular slot on T1 line till it is captured on particular card on inter working unit. After that the packet enters IP world.
So its a virtual circuit between MS – BTS and IWU. So why can’t i use Erlang B for calculating for the channels between BTS and IWU?
Anjum9th February 2006 at 17:46 #32398coleinGuest