Research On Synchronization And Demodulation Algorithms For UWB Communications Systems

Ultra-wideband impulse radio (UWB-IR) is a promising technology for data communications in short-range indoor applications. It is characterized by the transmission of ultra-short pulses (in the sub-nanosecond scale) at low power spectral density. In order to keep the signal spectrum within the Federal Communications Commission (FCC) mask and obtain adequate signal energy for reliable detection, a single information symbol is represented by several pulses, each located in its own frame. To accommodate multiple-access and smooth transmit spectrum, pseudo random timing-hoping (TH) spread codes or/and direct-sequence (DS) spread codes are often employed. Interest in UWB is motivated by several features including: ample multipath diversity, low-complexity baseband transceivers, a potentially large user capacity, and potential to overlay existing narrowband systems. To harness these benefits, however, UWB-IR faces several challenges, among which synchronization and demodulation are particularly critical at the receiver end.Synchronization is a performance-critical factor in most communication systems, especially in UWB transmission. It stems from the fact that the transmitted pulses are very narrow and have low power. The dense multipath channel, through which the pulses bearing information propagate, is unknown at the receiver during the synchronization phase. Moreover, the bit error rate (BER) of UWB is very sensitivity to mistiming, the timing errors as small as fractions of a nanosecond can seriously degrade the system performance. To cope with these challenges, a number of synchronization and demodulation algorithms have been proposed in this dissertation. Summarized as follows:Relying on the prior knowledge of the DS codes, a non-data aided synchronization algorithm is proposed for DS-UWB systems. It remains operational in practical UWB settings with unknown multipath propagation. With the aid of the integrate-and-dump (I&D) over a frame duration and peak-picking from the output of the DS codes matching filter, the proposed approach can achieve frame-level timing acquisition within one symbol duration. The maximum significantly differs from other values due to the pseudo random property of the DS codes. Therefore, the proposed scheme has high immunity to noise effects.A data-aided synchronization scheme dispensing with searching procedure is proposed in this dissertation. With the help of judicious training symbols and symbol-level samples, the energy values of the truncated tail and head of the received symbol waveform can be obtained. With the two energy values in hand, frame-level timing acquisition can be acquired in closed form. It is worth noting that with the perfect knowledge of the multi-path channel, the proposed scheme can achieve timing synchronization at any desirable resolution. The proposed algorithm only exploits symbol-level samples and without searching process, thus it turns out to not only speed up synchronization, but also enjoys low-complexity.Regarding demodulation, RAKE is the first receiver considered for UWB systems. Unfortunately, this classical approach will cause high re- ceiver complexity since a large number of fingers are required to estimate. Furthermor, RAKE is very sensitivity to mistiming, the timing errors as small as fractions of a nanosecond can significantly degrade the system performance. To bypass channel estimation and enhance the robustness to mistiming, the single reference UWB (SR-UWB) system is proposed. Relying on the unique signal structure of SR-UWB, two joint synchronization and demodulation algorithms are proposed without employing any training sequences (non-data aided). With the aid of overlap-add method and energy detection technique, the first approach can acquired timing acquisition even if in the presence of inter-frame interference (IFI) and inter-symbol interference (ISI). With the synchronization parameter being available, the demodulation template can be easily constructed. Taking advantage of the synchronization parameter and the demodulation template, the decision statistic, which is then passed through a sign detector to result in the symbol estimates, can be attained by using of the correlation operation. In the second algorithm, in terms of the unique signal structure and the prior knowledge of the TH codes, the objective function can be established over a short observation interval. Peak-picking the objective function, the synchronization parameter can be acquired. After timing acquisition, the samples used to achieve synchronization can be re-utilized for the demodulation process by a zero-crossing comparator. As a result, the implementation complexity is markedly reduced in comparison with the first approach.Based on the prior knowledge of the DS codes, a joint blind synchronization and demodulation scheme is developed for differential transmitted reference UWB (DTR-UWB) systems. The proposed approach can achieve frame-level synchronization with the help of frame-rate samples. Taking advantage of the periodicity of the DS spread codes, the frame-level synchronization can be carried out even in one symbol interval. On the other hand, after timing acquisition, these frame-rate samples can be re-utilized also for the demodulation by a zero-crossing comparator. Thus the acquisition time and the implementation complexity are reduced considerably. Meanwhile, the noise cross noise effect is alleviated because a wide-sense average operation has been executed in synchronization phase, which is responsible for the enhancement of the detection performance. It is also worth noting that the proposed approach can be employed in multi-user scenarios without any modification.By the special auto-correlation pattern of the proposed training sequence (4 bits Baker codes), a redundance-included demodulation template (RDT) is constructed from the received signal. When the RDT is available, two approaches can be selected to detect transmitted information symbols. One is to demodulate transmitted symbols directly by the RDT, which does not require timing acquisition, and thereby has considerably lower complexity. The alternative is to first carry out timing acquisition by invoking simple energy detection, which is employed to shear the RDT to obtain a non-redundance-included demodulation template (NRDT), and then demodulate transmitted symbols by the NRDT. Although the second approach has higher complexity (depending on the level of timing error desired), it can obtain better BER performance compared with the first one owing to canceling the redundant noise. Furthermore, the demodulation template is derived from the received signal, so both approaches can avoid the effect of pulse distortions and cope with the timing error.Simulations and theoretical analyses validate that the proposed approaches which have been stated above enjoy better performances in comparison with the similar ones in terms of acquisition probability, normalized mean square error (NMSE) and BER.