Research On Symbol Synchronization And Channel Estimation For Military-Burst OFDM Systems

With development of high speed data transmission in military communications, orthogonal frequency division multiplexing (OFDM) will play an important role dut to its high spectral efficiency and robustness to multipath fading channel. In military communications, systems appling OFDM to RF carrier frequency hopping (FH) and OFDM systems in burst mode can be used to achive anti-jamming capacity. These two kinds of systems are collectively called military-burst OFDM systems, which raise high requirement for symbol synchronization and channel estimation. From the utility point, we focus on symbol synchronization and channel estimation for military-burst OFDM system in this dissertation, laying emphasis on the robustness and efficiency of estimation algorithms. The main contributions of the dissertation are shown as follows.1. The statistic property of timing metric based on symmetric correlation is analyzed, which provides theory basis for algorithm improvement and performance analysis.Analysis shows that backward motion of timing position occures when the peak of timing metric is used to detect the desired timing position in multipath channel, where the first path has a small gain. The backward motion results in intersymbol interference and interchannel interference. Therefore, the peak of timing metric is not suitable for the synchronization method based on conjugated-symmetric preamble.2. For the preamble with repeated-conjugated-symmetric structure, a new symbol synchronization method using both delayed correlation and symmetric correlation is proposed. Moreover, a novel low-complexity symbol synchronization method is designed by utilizing the delayed product additionaly.A searching-window metric using delayed correlation is defined to serach the peak of the timing metric based on symmetric correlation. Then the first path which may not be the strongest one is being detecting with a fixed threshold. The fixed threshold is set by theoretical analysis, which needs neither computer simulation nor derivation from the computed timing metric. Therefore, the proposed symbol synchronization has accurate synchronization performance and fast synchronization procedure. Furthermore, a method using delayed correlation, symmetric correlation and delayed product is presented to reduce the computation complexity and maintain good synchronization performance simultaneously, which contributes to a fast synchronization.3. For the preamble weighted by pseudo noise (PN) sequence, statistic distribution of the conventional timing metric is analyzed, and an adaptive symbol synchronization method using delayed correlation based on cyclic prefix (CP) and correlation property of PN sequence is proposed. Moreover, a new preamble structure is designed.The timing metric based on delayed correlation can suppress interference caused by noise and data symbols, and the adaptive synchronization scheme can enhance the robustness to time-variant multipath channel. The proposed preamble has partial repetition structure, and only half samples were weighted by PN sequence. The novel structure is used to enhance the resistance to ISI, and construct an impulse-shaped timing metric. Therefore, the proposed preamble improves the synchronization performance.4. Channel estimation methods based on discrete Fourier transform (DFT) are improved to enchance the robustness to path delay of multipath channel.Two conventional methods based on zero-padding DFT/IDFT for channel interpolation, time domain method and transform domain method, are testified to be equivalent. Then a new zero-padding method is proposed to decrease the reconstruction error for channel impulse response samples. For data processing before zero-padding, analysis shows that asymmetric filter rather than conventional filter in the transform domain should be adopted, and then a simple method to set cutoff frequency is proposed. Moreover, to decrease the influence of energy leakage caused by non-sample-spaced path, noise power estimation in frequency domain based on preamble is used to set threshold for detecting the channel tap with large energy in the time domain, and a fixed detecting region is adopted. Simulation results show that the above methods can improve the channel estimation performance.