Adaptive Opportunistic Beamforming in Ricean Fading Channels - Adaptive Signal Processing in Wireless Communications - page 178

2.6

2.4

2.2

2

Mobile

1.8

1.6

Fixed

1.4

1.2

1

Equal time scheduling

0.8

0 5 10 15 20 25 30 35

Number of users

FIgure 7.2 Multiuser diversity gain in fixed and mobile environments. (Reproduced from
Viswanath et al., 2002. © 2002, IEEE. With permission.)

Rayleigh fading environment is larger than that under the AWGN environment. This
is due to multiuser diversity. Figure 7.2 gives more insight into how the dynamic range
and the variation rate of the channel fluctuation affect the benefits of multiuser diversity.
The figure demonstrates the total throughput of the downlink channel of the IS-856 sys-
tem under the following two environments: (1) fixed—users are stationary but there are
objects moving around them, and (2) mobile—users move at a walking speed (3 km/h).
Under both environments, the total throughput increases with the number of users, but
the increase is more impressive in the mobile case. This is because the dynamic range
and the variation rate of the channel fluctuation for the mobile case are larger than those
for the fixed case.
In order to exploit multiuser diversity, every user needs to estimate its own instanta-
neous SNR γ k ( t ) by a pilot signal and feed it back to the base station. On the other hand,
the base station needs to properly schedule the transmission to all the users and adapt
the data rate based on the instantaneous channel condition. This technique has been
adopted in some communications systems, such as IS-856 [18].
In practical communication systems, two important problems must be solved before
multiuser diversity is implemented: fairness and delay. If the fading statistics of the
users' channels are the same, transmitting data only to the user with the highest γ k ( t )
not only maximizes the throughput of the system, but also maximizes the throughput
of every user. In reality, however, the fading statistics are not symmetrical. For example,
some users may have better average SNRs, because they are closer to the base station.
Furthermore, practical systems usually have a delay constraint on every user. Thus, one
user cannot wait for too long until the user's instantaneous SNR becomes the highest
among all the users.