Stress test a typical wifidog router

[scarle]

We have seen some problems during podcamp montréal and are wondering what are the typical limits (theoritical and practical) for a standard combo (linksys wrtg or similiar)?

We are planning coverage of MTL Intl Game Summit and are trying to do some capacity and contingency planning.

[gbastien]

Theoretically, wifidog itself is not a problem, the router is. Many things influence the maximum available throughput:

1. the area covered (The signal gets attenuated with distance, so users far from the AP have slower rates and slow down the whole network)
2. the placement of the AP (any obstacle between the AP and the users may slow down the network)
3. the burstiness of the network (in a conference, it can be expected that during recesses, more people will transmit at the same time, augmenting collisions and thus the number of retransmits, and slow down the network)

For protocol like 802.11a or g, the maximum throughput is 54Mbps. In practice, it is good to expect at most 30 Mbps (with protocol overhead). That means about 20-30 active users per AP. Now, one must insure that the link behind those AP can handle the load.

In practice, we (ML and I) tried a few things, but with our limited capacity, we cannot say much more than there seems to be total denial of service for the fourth client who tries to connect, with the 3 others flooding incessantly the AP (which is hardly a real-life scenario).

Do you have more details wrt the Mtl Intl Game Summit? Will isf do the wifi coverage? Is so, we'll need to see what we can do to maximize coverage and performances for users.