Research On ICI Cancellation Algorithms In High Frequency OFDM System

OFDM (Orthogonal frequency division multiplexing) technology has been widely used in many fields as a MCM (Multi-Carrier Modulation) technology. Its most important advantage is the higher spectral efficiency, compared to the traditional frequency-division multiplexing technology. This is due to its use of the orthogonality between sub-carriers. This feature allows different sub-carrier signals overlapping in the spectrum without affecting their respective reception. However, this advantage results in a potential trouble to OFDM as well, namely ICI (Inter-Carrier Interference). With the continuous increasing requirements in communication, OFDM system moves towords to the one with more sub-carriers, higher carrier frequency and faster mobility. This will result in a narrower sub-carrier interval and worse ICI. Meanwhile, the orthogonality of OFDM systems has become very fragile and the impact brought by ICI will not be ignored.In this paper, a research on ICI cancellation algorithms in high frequency wireless OFDM system is carried out, considering the complexity limitations of hardwares in high frequency system.A most simple one-dimentional equalizer with low complexity has been previously proposed to suppress ICI for OFDM system over doubly-selective channels. This algorithm makes an average of the channel responses at all times within one OFDM symbol to replace the whole channel response. As a result, the channel is equivalent to a time-invarant channel. Therefore, it's not very effective in suppressing ICI becausing of ignoring channel's time-varying characteristic. The traditional frequency domain equalizations, such as LS, MMSE algorithms, use all the channel information of the channel frequency response matrix to restore the signal at the receiving end, which achieve excellent ICI suppressing ability with high complexity. Especially for large sub-carrier number N, the hardware requirements will be high in specific implementations. A low-complexity MMSE algorithm can reduce the system complexity to some extent, using the adjacent carrier distribution characteristics of ICI and replacing the channel frequency response matrix with a banded structure. The algorithm achieves a compromise between performance and complexity by adjusting the width factor of the banded structure. Some performance must be sacrificed due to discarding part of channel information. Based on mathematical modeling of OFDM basic framework and formulas deriving, a low complexity frequency-domain linear equalization algorithm based on pre-processing is proposed in this dissertation. The new algorithm first extracts a time-variant factor from every channel response during the OFDM symbol under linear time-variant channel. Consequently, one-dimensional filtering pre-processing in time-domain can be achieved at the receiver by using these factors. As for the time-variant channel after time-variant factor extraction, the new algorithm uses banded structure to approximate its frequency response matrix, which achieves low complexity frequency-domain linear equalization at the receiver.According to the LTV (Linear Time-Variant) channel characteristics, a new ICI cancellation algorithm with much lower complexity is proposed, based on CSC (Channel Segmentation cascade). The new algorithm devises an equivalent method which transfers linear time-variant channel into the cascading of time-invariant channel and time-variant channel with no delay, using channel segmentation. Then one-dimensional equalization can be achieved at the receiver to suppress ICI, avoiding calculating the frequency domain channel response matrix and its inverse matrix. This algorithm reduces the system complexity further significantly.The output SINR performance is compared between these mentioned algorithms using Matlab simulations. In addition, the impact on system performance and algorithms'ICI elimination capacity introduced by channel factors, such as the width factor of the banded structure, the number of channel multi-path, channel correlation coefficient, channel power delay profile -20dB width factor, is discussed through simulation results and performance analysis.At the last part, the whole work of the dissertation is outlined and the further research issues are discussed.