Research On Key Techniques In Demodulation For Filtered Multitone System

Filtered multitone modulation system owns an excellent resistance to frequency-selective fading channel by dividing transport channel into numerous parallel sub-channels, moreover, it’s orthogonality of sub-channels turn out to be robust in varied communication environment through the application of strict band limited prototype filters. Currently, FMT has been extensively researched and applied. Based on this case and the actual project of HF multitone, the thesis focuses on the key techniques involved in demodulation, such as efficient structure of modems, blind carrier frequency-offset(CFO) estimation, joint synchronization of timing and frequency and time domain channel estimation. The concrete research content and innovative achievements are as follows.1. This thesis discusses the principle of FMT system firstly, describes some common prototype filters and explains the influences of them on system performance. What’s more, two efficient structure of modem are designed based on polyphase filter and period superposition, respectively.2. In order to expand the estimation range and improve the estimation precision of blind carrier frequency-offset estimator, an algorithm based on twice oversampled in receiver was proposed which utilized the correlation function with small delay between odd-even sampled sequences to estimate CFO. it has an lager range CFO and without known prototype filter information, what’more, it has a robust performance of timing errors.3. Considered that multipath channel interference would cause serious degradation performance of existing CFO estimation algorithms, an algorithm based on spectrum characteristics was presented. Analysis indicates that junction point of adjacent sub-carrier spectrum was the valley point of spectrum fragment, new algorithm ultilized this characteristics to estimate the carrier frequency-offset. Moreover, to eliminate the influence of the channel spectrum zero under harsh channel, the article used divided-superimposed spectrum to estimate CFO.4. When constructing cycle training sequence, by the impact of tailing prototype filter, there was overlap between training sequence and transmitted data, ignoring this overlap part of data would bring peak platform of timing metric, but using this part would introduce unknown interference. Based on problems mentioned before, an improved algorithm employing variable sliding window was proposed, increased length of windows to reduce the width of peak platform as first step, then combined with moving average method to completely eliminate peak platform of timing metric.5. For that the pseudo noise training can’t estimate transmission delay of fractional multiple sample period, a fine timing synchronization module was designed. After coarse synchronization, the cross-correlation of received sequence and local sequence could be found out that the sub-peak and peak ratio is a funcation of fractional transmission delay, used mathematical relationship between ratio value and fractional delay, fine synchronization could be achieved.6. Focusing on time domain channel estimation techniques, in FMT system, the prototype filters usually have large order, causing symbols trailing overlap. This may introduce interference on pilots under multipath channel, in view of approximate the interference, an iterative method was presented. Simulation results show that the new algorithm can effectively improve the accuracy of channel estimation, but no significant increase complexity.