Research On Fast-Convolution Multi-Carrier Transmission

Consumers’ need for better experience of wireless network has motivated researches on 5G wireless communication system. In recent years, Filterbank Multicarrier (FBMC), Generalized Frequency Division Multiplexing (GFDM), Universal Filtered Multicarrier (UFMC), together with Fast Convolution Multicarrier (FCMC), have received much attention as the next new waveform, which is one of the key techniques of 5G. FCMC is highly configurable with relative low complexity, which deserves further research. In this thesis, several key issues on FCMC system are investigated.Firstly, we derive the vectorized formulation of FCMC transceiver for the first time. We start from formulating the FC version of Finite Impulse Response (FIR) filtering process in vectorized form. Next we extend this formulation to interpolation/decimation filtering, synthesis/analysis filterbank, and finally the FCMC transceiver, which paves the way for quantitative analysis. Based on this formulation, we propose a complete structure of FCMC system. Furthermore, FCMC’s performance in terms of calculation complexity, signal reconstruction error, power spectrum and Peak-to-Average Power Ratio (PAPR) are analyzed in detail. The advantages of FCMC system are given.Secondly, we focus on FCMC low-complexity equalization method based on the vectorized formulation. A one-tap frequency domain (FD) equalizer for FCMC is proposed. We start from the basics of Minimum Mean-Square Error (MMSE) equalization and develop a FC version of MMSE equalizer. It processes in stream rather than frame, which does not need Cyclic Prefix (CP). Moreover, it can be merged seamlessly in FCMC system. To reduce the complexity of equalization, we exploit the hidden nature of FCMC equivalent channel, especially the asymptotic behavior of Toeplitz matrices. By using the fact that Toeplitz matrices can be asymptotically diagonalized by Discrete Fourier Transform (DFT) matrix, the equalizer developed above can be simplified to one7tap FD equalizer when FFT size is large enough. Simulation results show that FCMC receiver with one-tap FD equalizer or MMSE equalizer has nearly the same Bit-Error-Rate (BER) performance, which is similar to MMSE equalization in CP-OFDM system.Thirdly, we focus on FCMC pilot design and channel estimation. A low-complexity FD channel esti-mation is proposed. Estimation using training sequence is investigated. We derive the Least-Square (LS) and MMSE estimator using FD and time domain (TD) pilot, respectively. TD pilot outperforms FD pilot because FD pilot will destroy the spectrum of modulated signal while TD pilot will not. Furthermore, noise-reduction process is developed to enhance performance. Matrix inversion is avoided by appropriate simplification. Thus we propose the FCMC low-complexity FD channel estimation method. Simulation results show that the degradation of estimation accuracy between low-complexity estimator and MMSE estimator is negligi-ble. Also, sub-band with larger bandwidth has better estimation performance. With this channel estimation, nearly the same BER performance can be achieved as that with the channel known at the receiver.