21.11.2012 Views

Wireless Future - Telenor

Wireless Future - Telenor

Wireless Future - Telenor

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

signals can be made on one side only, at the<br />

receiver, where the channel can be easily measured.<br />

If available, the transmitter weights could<br />

be used in a joint transmit-receive optimisation<br />

to maximise the wanted powers. The formulation<br />

in equations 4 and 5 is of course equivalent<br />

to the angular description above, but it leads to<br />

the more normal situation in Figure 2 where<br />

there is a multitude of rays, and there is no<br />

chance with a few elements to put many nulls in<br />

the radiation pattern. The key point is that it is<br />

not necessary to do so. The description in equations<br />

4 and 5 is still valid so it is still possible to<br />

choose the weight factors appropriately to separate<br />

the two data streams and increase the capacity.<br />

Physically we can explain the situation as<br />

putting the unwanted signal in a deep fade instead<br />

of a null in a radiation pattern, the result is<br />

the same. Note that the scattering from widely<br />

separated scatterers is a necessity for the increased<br />

capacity; a line-of-sight situation where<br />

the two antenna arrays see each other directly<br />

would only lead to one channel, although with a<br />

larger gain. This follows directly from Shannon<br />

where the capacity in bits/s/Hz for independent<br />

parallel channels is given by<br />

�<br />

C = N log2 1+ (6)<br />

where P is the power over noise, or SNR, divided<br />

equally over the channels. The expression<br />

grows with N, the more so the larger P is, but the<br />

beauty of the array solution is that antenna gain<br />

will also grow with N, so we end up with the<br />

following<br />

P<br />

�<br />

N<br />

C = N log 2 (1 + P) (7)<br />

Telektronikk 1.2001<br />

90<br />

120 60<br />

2<br />

150 1.5<br />

1<br />

0.5<br />

210<br />

240<br />

270<br />

300<br />

30<br />

Beam 1<br />

0<br />

330<br />

Beam 2<br />

Figure 1a A simple example with two scatterers widely<br />

separated. The antennas are multibeam antennas with<br />

two beams, the blue ones (beam 1) having a null in the<br />

radiation pattern in the direction of beam 2, and vice<br />

versa. The antennas only have two elements, so they cannot<br />

also have a maximum in the wanted direction. The system has<br />

doubled the capacity, since the transmissions take place at the<br />

same frequency<br />

Multibeam antenna<br />

with complex<br />

antenna weights<br />

W 1<br />

W 2<br />

Thus parallelism is a very promising technique,<br />

it can only be realised in scattering environments<br />

with wide angular spreads, and multiple antennas<br />

at both ends are needed.<br />

Although we have seen that only the receiver<br />

array has to know the channel, it is an advantage<br />

to let the transmitter array know the channel as<br />

well, which would increase the P in eq. (7). The<br />

information must in all cases be spread over the<br />

transmit elements, and considerable research has<br />

gone into finding optimum ways to do this using<br />

so-called space-time coding (Tarokh et al [8]).<br />

Foschini [1] and Winters [2] realised that this<br />

situation could also be created with closely<br />

spaced elements in a scattering environment,<br />

where the scatterers would act as angularly<br />

widely spread parasitic elements of the array.<br />

90<br />

120 60<br />

2<br />

150<br />

1.5<br />

1<br />

30<br />

0.5<br />

180<br />

0<br />

210<br />

240<br />

270<br />

300<br />

330<br />

Figure 1b Two independent<br />

data streams are sent out and<br />

received over both antennas,<br />

and the weights are chosen<br />

such that there is isolation<br />

between the two beams<br />

41

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

Saved successfully!

Ooh no, something went wrong!