- This topic has 7 replies, 1 voice, and was last updated 21 years, 7 months ago by
Alex Yuen.
-
AuthorPosts
-
30th January 2003 at 02:30 #35022
Uriel Oko
GuestBusy hour (in Erlangs) is the number of hours of call traffic there are during the busiest hour of operation of a telephone system. There is more than one way of collecting the data;
1) time consistent, where a week’s data is collected and specific peak period (8 to 9 am, for example) is averaged out for the week.
2) Bouncing busy hour where selectively, the busiest peak busy times are collected and the busiest peak is selected.Which method is the true one preferred for Erlang calculation?
4th February 2003 at 04:19 #35023Don McK
GuestThe method you choose is up to you. Method 1 tends to under rate the busy hour trafic figures as a peak time on one day may be low traffic for the corresponding hour of another day.
Method 2 overlooks the traffic trend for the rest of the sample. For example a relatively low traffic sample across the period can be masked by one high peak in ttraffic which may be caused by an abnormal event.
I prefer to find the highest daily BH then average them. More precisely, I try to take a long sample period and average the ten busiest hours.
The needs of your client need also be considered, for example would you bother using a Busy Hour calculation to determine trunk/circuit quantity if you were dimensioing an emergency service call centre? more likely you would determine the peak busy hour for a known sample and make sure you over-engineered the route!
Also be aware that BH calculations in this manner may mask actual calling pattern, for instance if all calls come in in the first 10 minutes of the hour sample the overall traffic may seem low, but the client may need to cater for the busy 10 minute period !
5th February 2003 at 00:23 #35024Uriel Oko
GuestDear Don McK,
Thanks for your response. Can Erlang Table B be used in either case with equal results? What correlation, if any, exists between the two methods?
Uriel
6th February 2003 at 00:17 #35025Don McK
GuestErlang B can be used for either case. Correlation is really non existant unless you can prove that the sample can be repeated accurately, many times.
Remember that this is a prediction, based upon numerous assumptions about the sample. It is improbable that all assumptions are in fact true for the (any) sample and therefor there is a ‘margin of error’ in the calculations.
None the less, it is my experience that Erlang B provides a very close estimate of the lines required
7th February 2003 at 00:26 #35026Don McK
GuestDon,
Thank you for your help. I truly appreciate it.
Sincerely,
Uriel Oko
7th February 2003 at 19:57 #35027Alex Yuen
GuestI am tele traffic engineer over 20 years and would like to give some information regarding to the term of busy hour traffic.
1. We need to be very careful when selecting the busy hour traffic. It may be different for local call traffic and long distance call traffic. The first thing is you need to find the actual busy hour for the service. It may be the actual busy hour traffic is much higher than the time consistant busy hour traffic. So for reality and for grade of service purpose for particular service, you need to find the actual busy hour. Do they have seasonal peak? such as for long distance call, they will have seasonal peak in Christmas and new year.
2. When you find the actual peak busy hour traffic, you need to plan for taking routine traffic measurement to define the time consistant busy hour traffic.
3. Normally we will use Erlang B for provisioning and dimensioning purpose.
4. Be careful that if you use VoIP, then some basic concept of traffic thoery will be changed.I worked in traffic engineering department over 20 years and experience in dimersioning of digital equipment and now is learning in VoIP.
9th February 2003 at 08:26 #35028Frando
GuestHow to calculate the busy hour trfiic if only the one day traffic volume data is available. Can I use the percentage of the one day period traffic volume to get the busy hour traffic? . If it is appropriate how many percent can i use to predict the busy hour traffic?
thankyou..
10th February 2003 at 23:07 #35029Alex Yuen
GuestIt is rare to use one day traffic volume data to dimensioning purpose. We need to know what kind of service that only can capture with one day traffic volume. I don’t recommend to use one day traffic. Usually we need to use five days average with continue of 12 months traffic records. But sometimes you can use only three days average.
Provisioning and dimensioning in telecom network is quite tough since everyone need to cut the cost. But we still need to provide professional & reasonable dimensioning method, the traffic measurment is most important in such aspects.If you still have questions related to traffic, you can directly contact me at alexkkyuen@ieee.org.
-
AuthorPosts
- The forum ‘Telecom Design’ is closed to new topics and replies.