04.02.2014 Views

Lecture Series in Mobile Telecommunications and Networks (1583KB)

Lecture Series in Mobile Telecommunications and Networks (1583KB)

Lecture Series in Mobile Telecommunications and Networks (1583KB)

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

As far as the rout<strong>in</strong>g of packets is concerned, it turns out that you can pretty much follow a straight l<strong>in</strong>e path, <strong>and</strong> such<br />

straight l<strong>in</strong>e paths will average out the load over the network so that there are no hotspots. That is an order optimal<br />

architecture. That is what we will get if we treat <strong>in</strong>terference as noise <strong>and</strong> there is a law of dim<strong>in</strong>ish<strong>in</strong>g returns, as I have<br />

po<strong>in</strong>ted out.<br />

But is spatial reuse the right design pr<strong>in</strong>ciple?<br />

better than deserts for wireless networks, if that is true.<br />

However, the fundamental question to ask is, is spatial reuse<br />

the right design pr<strong>in</strong>ciple? Why do I want to challenge<br />

that? As I po<strong>in</strong>ted out, spatial reuse, as we can see, is not<br />

used <strong>in</strong> an enabl<strong>in</strong>g cell but further away. If you really<br />

believe <strong>in</strong> spatial reuse of frequency, then you should<br />

believe that a sharper path loss is better for wireless<br />

networks. What do I mean? As radio signals travel, their<br />

amplitude attenuates. This is a gradual attenuation [on<br />

slide] <strong>and</strong> that is a more rapid attenuation. The philosophy<br />

beh<strong>in</strong>d spatial reuse is that if you do not want to have<br />

<strong>in</strong>terference from people further away then you should<br />

prefer a sharper attenuation. If you believe <strong>in</strong> that, then you<br />

should believe that this [on slide] is better than that. Or, to<br />

put it more colourfully, you should believe that jungles are<br />

Is spatial reuse really the right way to operate wireless networks?<br />

8 The Royal Academy of Eng<strong>in</strong>eer<strong>in</strong>g<br />

The problem is that wireless networks are formed by nodes<br />

with radius, not with wires. Therefore, to draw a picture like<br />

this is wrong, <strong>and</strong> it is rem<strong>in</strong>iscent of wires. Actually, you<br />

cannot see it here, but we have a whole bunch of antennae<br />

which are just radiat<strong>in</strong>g energy. That is the picture of a<br />

network that you should have, with everybody talk<strong>in</strong>g away.<br />

There is no a priori notion of l<strong>in</strong>ks – nodes simply radiate<br />

energy <strong>and</strong> so the network is Maxwellian rather than<br />

Kirchoff. In the Maxwellian world, strange th<strong>in</strong>gs can<br />

happen. Nodes can actually co-operate <strong>in</strong> much more<br />

complicated ways than they could <strong>in</strong> a wire-l<strong>in</strong>e network.<br />

For example, it turns out that if somebody shouts <strong>in</strong> your<br />

ear at the same time as somebody else is whisper<strong>in</strong>g softly,<br />

Shannon would actually say that was fantastic. In fact, the<br />

louder this person shouts, the better. Why? Because, when this person shouts really loudly, you can decode that person<br />

perfectly <strong>and</strong> subtract out that person’s signal if you know the channel activation, <strong>and</strong> pick up the whisper. It turns out<br />

that it is actually more difficult <strong>and</strong> you can decode neither of them – so that when one is louder, that is actually better<br />

for you. The po<strong>in</strong>t is that <strong>in</strong>terference is not <strong>in</strong>terference: everyth<strong>in</strong>g is <strong>in</strong>formation – even <strong>in</strong>terference – <strong>and</strong> the only<br />

question is how you deal with it.<br />

Here is another bizarre notion. We have this notion of signal to <strong>in</strong>terference plus noise ratio. Signal is the good guy <strong>and</strong><br />

<strong>in</strong>terference <strong>and</strong> noise are the bad guys, <strong>and</strong> this ratio tells you how good the signal is, compared to what is <strong>in</strong>terfer<strong>in</strong>g.<br />

Usually, we try to mitigate <strong>in</strong>terference but perhaps we can try to cancel <strong>in</strong>terference actively, just as <strong>in</strong> your acoustic<br />

sound-cancell<strong>in</strong>g headphones. For example, this node could transmit someth<strong>in</strong>g which cancels the effect of what this<br />

node is transmitt<strong>in</strong>g at this po<strong>in</strong>t, so you can have co-operative cancellation <strong>and</strong> perhaps you should try to reduce the<br />

denom<strong>in</strong>ator.<br />

Alternatively, perhaps you should not even buy <strong>in</strong>to the digital revolution. Instead of re-generat<strong>in</strong>g the packet at each<br />

<strong>in</strong>termediate node, why not just amplify what you heard, without bother<strong>in</strong>g to decode? In fact, the <strong>in</strong>formation is only<br />

<strong>in</strong>tended for the f<strong>in</strong>al dest<strong>in</strong>ation, so why should you <strong>in</strong>sist that <strong>in</strong>termediate nodes should be able to decode

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!