Orthogonal frequency division multiplexing (OFDM) is an attractive technique for wireless communications. However, in high-mobility scenarios, the time-variation of mobile radio channels over an OFDM symbol leads to a loss of subcarrier orthogonality, and resulting in intercarrier interference (ICI) which severely degrade the OFDM performance and introduce an irreducible error floor. In this thesis, a novel scheme is proposed to estimate the channel in OFDM systems. The key idea is to distort the data of OFDM symbol in frequency domain, such that an impulse signal is used to estimate the channel, in time domain at pilot samples. Then, a piecewise linear model is used to estimate the channel variation over an OFDM symbol. Simulation results show the proposed scheme can achieve a substantial improvement in the bit error rate (BER) performance of OFDM compared with Zhao, Chang, and Mostofi schemes. Moreover, the error floor significantly is reduced, particularly, at low signal to noise ratio (SNR) regions compared with the previously mentioned schemes. Recently, Mostofi proposed a channel estimation scheme to mitigate ICI in OFDM system by approximating the channel variation over OFDM symbol by piecewise linear model. But, for high Doppler spread the channel over OFDM symbol exhibit high order variation. Thus, a generalisation of Mostofi scheme is proposed, where a general polynomial model is used to estimate the channel. Simulation results show that at a high Doppler spread, the generalised scheme show remarkable improvement in the BER performance of OFDM over the Mostofi scheme. Additionally, in this thesis, a modified of “better than” raised cosine pulse-shape is proposed to improve the performance of OFDM in the presence of frequency offset. Simulation results demonstrate that the proposed pulse outperforms raised-cosine pulse and “better than” raised cosine pulse in terms of BER performance, ICI reduction and SIR enhancement.
展开▼
