Strange WiFi Behaviour

I am sure we are all familiar with WiFi networking, it has become the network of choice for many of us because of its ease of installation and a major benefit for the ladies amongst us is the fact there are no wires. I recently went to a client premises where the owner was insistent that we keep the wires to an absolute minimum, the hallway had just been decorated and the router to the internet (which our equipment has to connect to) was in a small building at the back of the premises (the office).

Netgear Wireless ExtenderThe router is intermittently detectable by some computers in the main building and is not detectable at all by our device which, naturally is located in the hallway. To overcome this problem we decided on a wireless extender (Netgear WN3000RP) the mobile phone we brought with us and placed half way between the main entrance and the office at the rear of the building was easily detectable all the time and we were able to connect to the internet through its portable hotspot. Though the packet loss was terrible.

17143 packets transmitted, 7732 received, 54% packet loss, time 17175690ms
rtt min/avg/max/mdev = 39.499/334.675/76724.521/974.901 ms, pipe 77

These figures are from the device over the internet to a server in our datacentre via a VPN, normally the packet loss from a reliable internet connection is well below 5%. The premises is located where mobile signal was choppy at best and so we thought nothing of it and proceeded to ignore the problem until we could obtain a wireless extender. Eventually we setup the extender and could access the internet through the fixed line internet connection, would you like to see the results of my ping?

2144 packets transmitted, 867 received, 59% packet loss, time 128589887ms
rtt min/avg/max/mdev = 29.103/861.861/61411.933/5532.798 ms, pipe 2

This is over a longer period but you can see the losses are very high, this was the same over a shorter period but I neglected to save the actual statistics. Imagine working over a ssh connection to that machine! I was pulling my hair out ;-) .

We changed channels, changed them again, and again, detected the other networks around us and tried to align with them… nothing worked. We have now come to the conclusion that we were experiencing a hidden node problem and we were unable to find a way to let them see each other.

In more detail we were unlucky enough to have many WiFi networks converging on our position on all different channels. Not to mention the various other electronic devices, such as telephones, that may or may not have been utilising the 2.4Ghz band.

A lot like trying to talk with girls ;-)You will have to forgive the crude drawing but being in the edge of any of the circles where the signals intersect will be problematic as many frames will collide. Not forgetting that differing weather and other environmental factors can expand and contract these circles (oh, and they are only circles in the perfect case, any obstruction deforms the circle). For a better diagram have a look on Wikipedia, I promise my next diagram will be created with Inkscape ;-) .

In this configuration you can see that the collision avoidance techniques employed by the WiFi standards have not been successful and it turns out that the mobile network was probably fine it was the link from our computer to the portable hotspot causing the packet loss we saw earlier.

This problem can be solved in only one practical way… run a cable. You will also benefit from faster speeds as most network interfaces are gigabit and even though they will probably be limited to 100Mbps by your router it will very likely be quicker and more reliable than your WiFi connection.

http://lists.samba.org/archive/wireless/2002-January/000792.html

Comments are closed.