Link enhancement techniques for future multicarrier systems

Orthogonal Frequency Division Multiplexing (OFDM) is very effective in combating the distortive effects of wireless channel and promises high data rate capabilities with reasonable complexity and accuracy. Other advantages of OFDM include significantly simple equalizer requirement, high spectral efficiency, computationally inexpensive implementation, increased immunity to impulse noise, ability to support adaptive modulation schemes and high flexibility in resource allocation. This thesis investigates two vital issues regarding the OFDM system design requirements, namely, Channel Estimation (CE) and Carrier Frequency Offset Estimation (CFOE). The accuracy of these two estimators plays a crucial role in the overall performance of OFDM systems. Whether it is a single antenna or a multi-antenna OFDM system, accurate channel estimation (CE) is required for coherent reception. The channel estimation requirement is further exacerbated in the case of OFDM systems with multiple transmit and/or receive antennas as the signals are simultaneously transmitted / received and consequently arrive at the receiver through many channels. CE techniques for OFDM systems are broadly classified into pilot-aided and blind techniques. As compared to blind algorithms, pilot-aided CE algorithms are more robust to high Doppler frequency, and hence, are useful for high mobility applications. One of the major drawbacks of OFDM is its sensitivity to time and frequency synchronization errors. Owing to its inherent cyclic symbol structure, the time synchronization requirements are somewhat relaxed for OFDM systems. Conversely, the frequency synchronization requirements are more stringent because of its tightly packed subcarriers. The frequency offset results in loss of orthogonality of subcarriers which subsequently causes significant performance degradation. Therefore, it is imperative to estimate the CFO and thereafter eliminate or minimize its effects. This thesis proposes a new set of techniques for pilot-aided CE namely “undersampledchannel estimation” for OFDM systems. In such techniques, the number of pilots used to sample the channel are less than those allowed by Nyquist sampling theorem. Virtually blind (VB) CE uses only one pilot to estimate the channel frequency response (CFR). The performance of VB CE is hindered by the occurrence of CFR inversion (CFRI). Uniformly spaced fixed additional pilots and dynamically assigned additional pilots were then augmented with the only pilot in order to take more samples of channel and to stop propagating CFRIeffect further. For joint blind channel and control signal estimation for OFDM systems, the detectability of control information dependent (CID) pilot sequences is highly dependent on the type of sequences used. An algorithm to design a new set of pilot sequences with better detectability is proposed in this thesis.