Traffic Modeling for Wireless Broadband Services

[Presley]

We are hoping to find a closed, small to medium, campus environment whose Internet traffic statistics could be used to parallel the potential usage of a rural community in any state USA.

The objective is to determine daily Internet traffic statistics relative to the number of seats occupied and the amount of broadband pipe required to support said seats, factored by the ratio of occupied seats to the total subscriber base.

We feel this traffic parallel would be helpful in building a business case for deploying Broadband Internet Services in the rural communities of America. Wireless Broadband appears to be one of the cheapest and quickest means to do so.

Sighting the closed campus environment as a parallel, the Wireless Internet Service Provider (WISP) can determine how much carrier pipe is needed to support a given supscriber base, at various classes-of-services, and support a suitable ROI to fund the project.

If anyone can direct us to a site with traffic statistics that factor in the seat scenario, it would be greatly appreciated.


Presley...

 

[mainepe]

Suggest you also post this query to the Westbay Engineers erlang.com website <

http://www.erlang.com>

if you have not already done so.

In addition, you may find this article from the January, 2001 IEEE Communications Society Magazine interesting--
Traffic Theory and the Internet
Jim W. Roberts
France Telecom R&D
pp 94 - 99
The article offers up innovative paradigms for considering Internet traffic that enable the application of traditional traffic engineering theory to be applied to the traffic analysis and modeling of the Internet.

 

[mainepe]

I should have gone on to say that if one accepts the concepts espoused by the article, it would appear that one might not need to factor subscriber base data by occupied seats, nor need empirical data to enable the business case to be made.

This would then reduce your analysis down to a function of the projected average length and duration of transaction, and the total subscriber base. In essence, not much more than that which is done to project the number of trunks needed to support a call center with a given number of agents, where average call duration is known, and design grade of service is prespecified.