Estimation and Performance Analysis of Wireless Multiple Antenna Communication Channels
The focus of this thesis is on channel estimation in wireless communication systems that employ multiple antennas.
The thesis is divided into two parts, whereof the first part addresses the
problem of parameter estimation of distributed sources. Due to, e.g., local
scattering around the transmitter, the source (as seen from the receiver)
appears spatially distributed. A characterization of the spatial channel, in
particular mean direction of arrival and spatial spread, is of interest for optimization
and performance prediction of future communication systems. Nonparametric
beamforming-based estimators are derived and analyzed for the
localization of distributed sources, and it is found that they provide good
estimation performance. Parametric generalized beamformers are also presented
and analyzed. It is found that they also yield competitive performance
compared to existing algorithms.
The second part of the thesis deals with Multiple-Input Multiple-Output
(MIMO) and Single-Input Single-Output (SISO) channel estimation. A simple
and straightforward way of estimating the unknown channel is to transmit
known training/pilot sequences. In the recent past, a number of publications
have suggested Superimposed Pilots (SIP) for channel estimation in communication
systems. However, the performance gain achieved by SIP compared
to conventional (time-multiplexed) training is still questionable. To evaluate
the performance of the various training-based schemes, a lower bound on the mutual information of a general training-based scheme applied to block-wise fading MIMO and SISO channels is derived and maximized. It is found that in certain scenarios it is beneficial to also transmit data during the training mode (i.e., use SIP). The main conclusion though, is that the general SIP-scheme quite often reduces to the conventional time-multiplexed scheme, and, hence, renders the same performance. The theory is also extended to the case when detected data symbols are used as additional training symbols, which significantly improves the channel estimation performance. An improved channel estimate leads to an improvement in the effective Signa-lto-Noise Ratio (SNR), and, thus, a higher coding rate may successfully be applied. The extended scheme shows a performance close to the fundamental capacity of the noncoherent MIMO channel.