You Have To Share The WiFi Bandwidth

The most common networking medium today is Ethernet. The most popular Ethernet uses 4 wires, 2 for sending and 2 for receiving, to provide 100M full duplex bandwidth. The equivalent to 100M Ethernet is 802.11g WiFi, which provides 54M half duplex bandwidth.

If you have just 2 computers with Ethernet adapters, the simplest thing to do is to connect both with a cross-over cable. If you have 3 or more computers, you’ll likely get a switch or router, and connect each computer to that, one Ethernet cable / computer. With full duplex switched Ethernet, you’ll get a total of 200M bandwidth in each conversation between a pair of computers – 100M sending, and 100M receiving. As you add computers and Ethernet cables, the total bandwidth provided by your network grows. This is why we say that an Ethernet network is scalable.

Wifi, on the other hand, is not scalable. With your computers connected thru WiFi adapters, whether directly to each other (ad-hoc mode), or to a WiFi router (infrastructure mode), all computers must use the channel together. No matter how many computers you have – 2, 3, or more, your computers will have to share the channel. And if your neighbour has a WiFi LAN on that channel, your computers will have to share the channel with your neighbours WiFi LAN.

By saying “share the channel”, I am saying that, when your WiFi router is transmitting, no other computer or router within range of your router can transmit. Only one device – computer or router – can transmit over any channel at any time.

To share the channel, a WiFi device uses a strategy called Carrier Sense Multiple Access/Collision Aviodance (CSMA/CA). CSMA/CA, which is similar to a strategy previously used by classical (pre-switched) Ethernet, is not an efficient strategy.

  • Each WiFi component has to listen to the channel for some amount of time, before transmitting, to ensure that nothing else is currently transmitting. Precious portions of your 11M (54M, 128M) bandwidth are wasted, when listening.
  • Even with each WiFi component listening to the channel before transmitting, it’s always possible to have a collision, when two or more components pick the same time to start transmitting. When there’s a collision, both components will have to retransmit; more of your bandwidth is wasted, when retransmitting.

With Ethernet, if you use the proper equipment and design your network within limits (mainly, with each computer connected, by no more than 100 metres of Cat-5 or better cable, to the router or switch), you’re pretty much guaranteed 100M bandwidth. With WiFi and CSMA/CA, the general estimate is that you will get 1/3 – 1/2 of the stated bandwidth. And that only involves your computers and router, with your router managing the relationship. When your neighbour’s WiFi LAN becomes involved (and both routers have to manage a peer-peer relationship), your channel availability, and bandwidth, drops further.

There are 11 802.11b channels, each capable of providing up to 11M of bandwidth (the maximum again). Using 802.11g, we get 3 channels, each capable of providing up to 54M of bandwidth.

802.11b    802.11g
1 - 3      Bottom ("1")
4          Empty
5 - 7      Middle ("6")
8          Empty
9 - 11     Top    ("11")

Now, 802.11b and 802.11g are ratified standards. Each manufacturer of standard equipment designs it to perform in a predictable way, so if your WiFi router has to share the channel with a router made by another manufacturer, it will perform properly. But 802.11g doesn’t provide enough bandwidth, so the manufacturers are now working on a new standard, 802.11n. The new standard includes 2 features (using names which vary by vendor):

  • MIMO.
  • Super-G.

MIMO, or Multiple-input Multiple-output, uses multiple radios and antennas. MIMO has two components.

  • Antenna diversity. If you’re familiar with FM radio in your car, and multi-path interference, you’ll know the value of antenna diversity. The idea behind antenna diversity is that, if the signal from a radio transmitter is weak on one antenna, because of MPI, it will, hopefully, be stronger on another antenna some distance away from the first. A special processor does nothing but compare the signal being received by two different antennas, and select the stronger.
  • Beamforming. Antenna diversity counter acts multi-path interference. Beamforming uses the principle of multi-path interference, at the transmitter, to focus the strength of the transmitted signal in one direction. Using the diversity antennas on a MIMO component, it’s possible to identify the relative location of the other device in communication; using beamforming, the transmitted signal is focused in that direction.
  • By combining antenna diversity and beamforming, it’s possible to extend the effective range of a WiFi conversation.

To get 128M, aka Super-G, all 11 802.11b channels are combined. There is one channel – “6”.

Both MIMO and Super-G will give you more bandwidth, and more effective range, assuming that you have no neighbours with WiFi. If you have neighbours (and who doesn’t), only one of you can use a channel at any given time. Your equipment will have to decide how to share the channel. But, there are additional issues here.

  • MIMO will increase the effective size (area) of your WiFi neighbourhood, by increasing the effective distance between WiFi components that can receive each others signals. This increases the number of devices that have to share the channel, at any time.
  • Super-G will increase the size (volume) of your WiFi neighbourhood, by using more of the frequency spectrum to create more bandwidth. More channels used by your WiFi router increases the number of devices that have to share the channel, at any time.
  • More devices that have to share the channel means less time each device can transmit, and less bandwidth available to each client device. More devices that have to share the channel means more possibility of collisions, at any time, and again, less bandwidth available to each client device.

The dynamic effect of MIMO beamforming may have another effect. When you setup a WiFi LAN, you’re advised to try different channels (most effectively, using NetStumbler or a similar site survey tool). Over some period of time, you should be able to identify the majority of your WiFi neighbours, and pick a less congested channel. With beamforming, you’ll have a dynamic signal pattern, which will change as a WiFi client is moved around the house. There will be constantly changing visibility of WiFi neighbours, on any given channel (or group of channels). This will cause problems similar to the WiFi hidden node problem.

Think of being in a small room, with 2 or 3 of your friends. You can converse with no problem. Now, take away the walls of the room, so you’re all in the middle of a large room with dozens of neighbours. How easily can you converse then?

As you increase the effective size (area / volume) of your WiFi neighbourhood, your WiFi components will be able to detect (“see”) more WiFi networks using any channel. Since only one WiFi device can transmit at any time, your WiFi network will spend more time waiting to use the channel. When simply waiting becomes unsuccessful, it will spend additional time recovering from collisions. More waiting / collisions = less effective bandwidth = slower file transfers. Pure and simple.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: