Advanced interference management techniques for future wireless networks
View/ Open
Date
30/06/2014Author
Razavi, Seyed Morteza
Metadata
Abstract
In this thesis, we design advanced interference management techniques for future wireless
networks under the availability of perfect and imperfect channel state information
(CSI). We do so by considering a generalized imperfect CSI model where the variance of
the channel estimation error depends on the signal-to-noise ratio (SNR).
First, we analyze the performance of standard linear precoders, namely channel inversion
(CI) and regularized CI (RCI), in downlink of cellular networks by deriving the
received signal-to-interference-plus-noise ratio (SINR) of each user subject to both perfect
and imperfect CSI. In this case, novel bounds on the asymptotic performance of linear precoders
are derived, which determine howmuch accurate CSI should be to achieve a certain
quality of service (QoS). By relying on the knowledge of error variance in advance, we
propose an adaptive RCI technique to further improve the performance of standard RCI
subject to CSI mismatch.
We further consider transmit-power efficient design of wireless cellular networks. We
propose two novel linear precoding techniques which can notably decrease the deployed
power at transmit side in order to secure the same average output SINR at each user compared
to standard linear precoders like CI and RCI.
We also address a more sophisticated interference scenario, i.e., wireless interference
networks, wherein each of the K transmitters communicates with its corresponding receiver
while causing interference to the others. The most representative interference
management technique in this case is interference alignment (IA). Unlike standard techniques
like time division multiple access (TDMA) and frequency division multiple access
(FDMA) where the achievable degrees of freedom (DoF) is one, with IA, the achievable
DoF scales up with the number of users. Therefore, in this thesis, we quantify the
asymptotic performance of IA under a generalized CSI mismatch model by deriving novel
bounds on asymptotic mean loss in sum rate and the achievable DoF. We also propose
novel least squares (LS) and minimum mean square error (MMSE) based IA techniques
which are able to outperform standard IA schemes under perfect and imperfect CSI. Furthermore,
we consider the implementation of IA in coordinated networks which enable us
to decrease the number of deployed antennas in order to secure the same achievable DoF
compared to standard IA techniques.