 Abstract | Orthogonal Frequency-Division Multiplexing (OFDM) is now regarded as a feasible alternative to the conventional single carrier modulation
techniques for high data rate communication systems, mainly because of its inherent equalisation simplicity. Transmitter diversity can effectively
combat multipath channel impairments due to the dispersive wireless channel that can cause deep fades in some subchannels. The
combination of the two techniques, OFDM and transmitter diversity, can further enhance the data rates in a frequency-selective fading
environment. However, this enhancement requires accurate and computationally efficient channel state information when coherent detection is
involved. A good choice for high accuracy channel estimation is the linear minimum mean-squared error (LMMSE) technique, but it requires a
large number of processing operations. In this thesis, a deep and thorough study is carried out, based on the mathematical analysis and
simulations in MATLAB, to find new and effective channel estimation methods for OFDM in a transmit diversity environment. As a result, three
novel LMMSE based channel estimation algorithms are evaluated: real time LMMSE, LMMSE by significant weight catching (SWC) and low
complexity LMMSE with power delay profile approximation as uniform. The new techniques and their combinations can significantly reduce the
full LMMSE processor complexity, by 50% or more, when the estimation accuracy loss remains within 1-2 dB over a wide range of channel
delay spreads and signal-to-noise ratios (SNR). To further enhance the channel estimator performance, pilot symbol structures are investigated
and methods for statistical parameter estimation in real time are also presented. |
