, an Italian inventor, proved the feasibility of radio communication. He sent and received his first radio signal in Italy in 1895. At first spark gap transmitters were used, transmitting in every direction at once, and detectable everywhere within range. There was one channel, more or less, worldwide.
Tunable circuits then were used. This meant that different bands were opened up, the bands used for FM, AM, LW etc were independently licensed off to the different broadcasters and users of the spectrum. There were thousands of channels worldwide, and millions of satisfied users. These bands were set for 1900s technology. How many users can we squeeze in with 21st century tech? Are we anywhere near reaching a limit?
That's a huge question. Ok, this node just examines how much capacity the higher frequencies that are being used for WiFi can support.
The main techniques to improve capacity in a radio network are as follows:
- tuning to different frequencies
- directional antennas
- transmitting using minimum power
- node routing
Ok let's go through them one by one:
This uses a different carrier frequency for each channel. This means that the different signals are separable at the receiver and thus the capacity is proportional to the number of channels that are defined multiplied by their width. There's a limit to how many channels you can have though, because if the channels are too close they interfere with each other.
Using directional antennas, transmitters and receivers can multiply up the capacity of wireless networks. Basically, provided two transmitters are at a different angle to the receiver then they may both use the same frequency. This usually multiplies up the amount of bandwidth available by the gain of the aerial; some aerials have 100:1 gain, which would give upto 100x more capacity. However aerials that big tend to be bulky, but smaller gains are easy and compact. Note that a directional aerial does not need to move mechanically, they can be electronically steered, you don't have to wear it on your hat ;-)
The idea behind this is that no transmitter should transmit more intensity than it needs to, the more power it sends, the further the range of the signal will be. By minimising the power, other people using the same frequencies can be closer, and there's less noise. Also, as a side benefit, if the equipment is battery powered, they are going to last longer.
If I want to send a message around the world I could either adjust my radio up to the megawatt range and blast it there, taking up a whole channel worldwide or else I could send a low power message to fred, who sends it to jim who sends it to john who ... eventually it gets there. It sounds more complicated, but note that the capacity of the network is much greater. By avoiding monopolising the whole channel worldwide I have allowed billions of other messages to use the same channel. Therefore node routing increases the available bandwidth, I've used 'n' hops worth of bandwidth but saved n^2 hops by cutting back power.
For a long time 'multipath distortion', (i.e. echos off neighbouring objects) was considered a real nuisance. Recently it has been realised that it's actually a great way to have multiple radios working on the same frequencies. By recording the signature of the echos modern receivers can use that as a signature of where the transmitter you want to listen to is, and filter out the ones you don't. It massively increases the capacity of the network because many, many users can all use exactly the same frequencies.
OK! Put all 5 together and what do we have?
Well, these techniques are synergistic- each one enables greater usage of the others. Studies have shown that if you do ALL this, the total available bandwidth of the whole network is proportional to the number of computers. That means that the more computers there are on the network, the bandwidth each has stays constant. In some ways it's quite a surprising result; normally with radio, more users means less bandwidth each. Here, more users mean no change, it all scales up.
In a rough way of speaking- the bandwidth of wireless networks is almost infinite.
- I thought that capacity was limited by Shannon's Law?
Yes, it is, kinda. Shannon's law applies to wireless point to point links. It doesn't directly describe the whole network though, because the total noise depends on how you send, what frequency you send and what direction you send in. These are not considered or constrained by the Shannon equation, so we can optimise them to build a wireless network.
- does this work for any frequency band?
To some extent, but it works better at high frequencies. The maths only works out in dispersive medium where the signal gets blocked by terrain, buildings and such like. Then the capacity is much larger- it's much the same as it is easier to have a conversation in a house with lots of sound deadening than a cathedral that echoes.
- What about the noise floor caused by a huge network?
The microwaves get absorbed by the surroundings, get lost due to curvature of the earth or absorbed and ignored at the receivers. This means that the range is limited for each transmitter- the signal goes down faster than inverse square law with distance; this means that the contribution of very distant transmitters can be ignored- they make no difference; and this limits the noise floor.