Adaptive Otdm Systems Based On The Optimal Signal Subspace

With broadband communication's development, block-wise transmission techniques have become a hot research field. There are two CP-aided block-wise transmission techniques, OFDM and SC-FDE. OFDM, featuring high frequency efficiency, robustness to frequency selective fading and simple equalization, is sensitive to carrier offset and has peak-to-average power ratio problem; SC-FDE, as is called OTDM herein, is different from conventional single carrier techniques and transmits data block-wisely. By use of CP it can also adopt simple frequency domain equalization methods, additionally it has the advantage of low PAPR over OFDM.Firstly, this thesis introduces the temporal-frequency duality of OFDM and OTDM. Current adaptive techniques in OFDM are analyzed, including bit-loading and pre-equalization, and it is pointed out that there exists no such technique in OTDM until very recently. Further, the two famous linear frequency domain equalization methods in OTDM is discussed and the conclusion is drawn that it is critical for OTDM to maintain signal wave form un-distorted and noise power un-amplified after equalization.Secondly, this thesis introduces several original concepts, called temporal symbol, frequency symbol, temporal domain subchannel and frequency domain subchannel, on which Optimal Signal Subspace theory is established. By this means, the receiver can determine an appropriate OSS, forms corresponding CSI for the transmitter to utilize. This is a low-cost adaptive scheme that needs only one bit feedback information per each subchannel. Observing that the OSS dimension is a critical parameter that links system power and spectral efficiency, we also provide two OSS dimension determination criteria, i.e., Gaussian criterion and channel capacity criterion. This proposed OSS-based adaptive OTDM system is simulated over multi-path channels, concerning the two criteria respectively. Simulation results show that system BER performance has high consistency with mathematical deduction. The PAPR problem in this system is also analyzed, and a method of dividing non-2's power-point DFT into sub-DFT-blocks is provided to reduce system computational complexity.