Performance Comparison Between OFDM And Wavelet (Lifting) Multi-Channel Modulation

Digital signal processing is now playing more and more important role in communications with the evolution of digital communication systems and highly improvement in VLSI technology. Traditionally, DSP only concerns in source coding as well as channel coding, but today, it begins to set foot in radio frequency modulation field. Now, many researches focus on baseband modulation in which digital filter banks are employed.Wavelet, as an alternative analysis method for classic Fourier Transform, has earned many attentions recently. That lifting-based wavelet transforms adopted by JPEG 2000 indicates wavelet's success in source coding. The motivation of this thesis is to extend wavelets' application to modulation field. In fact, our wavelet multi-channel modulation (WMCM) system constructed by substituting DFT with discrete wavelet transform, from DSP perspective, is a parallel transmission system which baseband modulation implemented by recursion of two-channel filter banks.In the thesis, algorithmic complexity, peak-to-average power ratio (PAPR), frequency shift and pulse interference are all involved in performance comparison between OFDM and WMCM, as well as bit error rate (BER) under AWGN and typical wireless channel. As it is to OFDM, WMCM has lower time complexity in baseband modulation, good PAPR distribution, and takes on more robust to pulse interference. But both of them are sensitive to carrier frequency shift, further, they have comparative BER under AWGN and Doppler shift environment. As OFDM adopts cyclic prefix, which transfers channel's effect from convolution to multiplication, it is easy for the system to reach performance under AWGN with proper channel estimation while multi-path propagation exits. Unfortunately, there is not this simple mapping in wavelet, so it is hard to eliminate multi-path effect for WMCM system. Both orthogonal and bi-orthogonal wavelets are employed in our research, generally, system based on orthogonal wavelets are more superior to those based on biorthogonal ones.