Physical Layer Parameter and Algorithm Study in a Downlink OFDM-LTE Context

This Ph.D. is made in cooperation between Infineon Technologies Denmark, and the Radio Access Technologies (RATE) section at the Department of Electronic Systems. The development of the LTE standard has been based on new increased requirements, with high demands for spectral efficiency, reduced system and terminal complexity, cost and power consumption. This leads to investigate the duality between physical layer parameter and baseband receiver algorithm design in this Ph.D. thesis. More specifically the work has been focused on physical layer parameters and baseband algorithms design for OFDM in a downlink LTE context. The study is based on an accurate baseband matrix-vector model of the received signal. The model is useful to separate two different cases: a Cyclic Prefix (CP) length larger than the radio channel maximum excess delay, and a CP shorter than the maximum excess delay. The LTE baseband parameter design has been investigated with an emphasis on optimal CP length. In the case of a CP length larger than the maximum excess delay, an in-depth survey of linear Pilot Assisted Channel Estimation (PACE) algorithms has been conducted leading to the development of a novel unified modeling for fair comparison of PACE algorithms. The effect of virtual subcarriers as well as non-sample-spaced channel model is studied, showing that DFT based algorithms are only useful at low SNR values, but are subject to the leakage effect as the SNR increases (above 10dB). A clear dependency between a-priori information considered at the receiver and performance is established. To avoid the leakage effect only two algorithm types are useful: robust wiener filters v and interpolators using exact channel tap delay knowledge. A methodology for pilot pattern design has lead to a pilot scheme proposal for the downlink of LTE. Results are characterized without error correction coding in terms of uncoded BER, SINR and mean squared error estimates, and with error correction coding in terms of packet error rate and spectral efficiency. In the case of a CP length shorter than the maximum excess delay, interference cancelation techniques were investigated to cope with insufficient guard interval length, leading to the development of a novel algorithm: the Low-Complex-Interference-Cancelation (LCIC) algorithm. Overall in this work, it has been shown that signal processing effort spent in the UE can increase the system spectral efficiency. If effort is spent on accurate tap delay estimation, much lower frequency direction pilot spacing can be used. In the same manner, the CP length can be reduced by using the proposed interference cancelation schemes.