ND SatCom SKYWAN IDU 1070, Manual

Page: 49 / 154

2010-10-26 Network Design and Engineering Guide 45
General Carrier Design
Network Traffic Estimation
quirement is as follows: Two bidirectional video conferences with 256 kbps per direction should
be simultaneously possible in the network. Each conference should be set up between one
head office (type 1 station) and one small site (type 4 station). The assignment of 256 kbps for
each station of type 1 is still correct. For the type 4 stations the situation is more complex: As-
signing 256 kbps for each type 4 station would lead to a network real-time traffic requirement
of 42*256 = 10752 kbps which is definitely too high. On the other hand, spreading the real-time
capacity requirement for the type 4 stations of 512 kbps among all small sites by defining only
512/42 = 12.2 kbps for each station would lead to the correct network traffic requirement.
However, if not all stations of this type are members of the same downlink population, this still
could lead to wrong results. Assuming, for example, that one downlink population consists of
only five type 4 stations, the required real-time capacity for the carrier of this population would
be estimated to be 61 kbps which is even for one video conference not sufficient. For this rea-
son it is more reasonable to assign the real-time capacity for the type 4 stations not to individual
stations but to the whole network as it is displayed in the following picture:
Figure 2-27 Per Network Traffic
Erlang B Calculation Worksheet
The second work sheet includes formulas for an estimation of required voice circuits using the
Erlang B distribution. This statistical method allows the calculation of the blocking probability
(i.e. rejected calls due to network congestion) for a network with a given number of users, an
average and peak value for the user’s hold time and the available number of voice circuits. Note
that Erlang calculations provide accurate results only for sufficiently large networks (more than
50 users). An example of such a calculation is presented in the following picture. The assump-
tions for this example are:
- Number of users (identical to the number of voice interfaces): 200
- Service hours: 10
- Average hold time: 12 min/hour
- Peak factor = 2, meaning that in the most busy hour the telephone is used twice as often
than on average
- Maximum acceptable blocking probability: 1.5%
Whereas the first four items are input parameter in the Erlang B worksheet, the blocking prob-
ability is derived from these input parameters and the number of voice channels, which is an-
other input parameter of the sheet.
To determine the required number of voice channels in the network, an initial estimate has to
be done and the resulting blocking probability observed. If the resulting blocking probability is