Research On Synchronization Based On Spectral Estimation And Cyclic Autocorrelation For UWB Systems

Ultra-Wideband (UWB) technology has received increasing attention especially in the area of short-range indoor wireless communications. It originates from the unique advantages of UWB signals: avoided symbol interference due to the low-duty-cycle pulse; high user capacity with time hopping (TH) codes; high data rates along with low transmission power; low probability of detection; enhanced capability to penetrate obstacles; ultra high precision ranging at the centimeter level; and potential spectrum compatibility with legacy services. However, UWB faces several challenges among which low-complexity and high-accuracy synchronization algorithms are particularly crucial in multipath indoor environments. This dissertation offers an in-depth study for UWB synchronization, and proposes three low-complexity spectral-estimation-based algorithms and a chip-level synchronization scheme as follows:Relying on the property of data matrix, a subspace-based synchronization algorithm is proposed firstly. It obtains the required signal subspace by carrying out linear operation for the data matrix without singular value decomposition (SVD), and estimates synchronization parameter by exploiting the shift invariant-subspace property of the signal subspace. Compared with conventional methods, the proposed scheme has lower computational requirements because SVD is avoided, and has well performance.By constructing the reasonable transmitted model, a data-aided algorithm is proposed in frequency domain. The synchronization parameter and nuisance ones are separated from each other on account of the structure of the Vandermonde matrix, and their estimates are obtained by one-dimensional searching according to the orthogonality among vectors. During the searching process, relative path delays and attenuations come out as by-products. Most importantly, the proposed algorithm has a very low complexity, because the computational requirements are determined mainly by simple additions and inner products among vectors. Simulations and comparisons illustrate the promising performance of the proposed algorithm.Using the first order statistics of the received signal, a low-complexity blind algorithm is carried out without aided data. It is similar to the second method, the synchronization parameter and the nuisance ones are distinguished from each other relying on the nature of the Vandermonde matrix, and estimated according to the orthogonality among vectors; channel parameters also come out as products. Most significantly, the computational requirements of this scheme are same with that of the data-aided one which above mentioned under the same condition, and the synchronization parameter is estimated without losing of transmission efficiency. For better performance, supplemented additions are involved, but only a neglectable complexity is appended to this algorithm. Simulations show that the proposed algorithm is very applicable for UWB synchronization.Employing the cyclic autocorrelation of the spread time hopping (STH) codes, a novel scheme is developed for chip-level synchronization. To accomplish the proposed algorithm, TH codes is transformed to STH codes firstly, without changing traditional modulation mode of UWB. Invoking the cyclic autocorrelation of STH codes, the synchronization parameter is acquired without training sequence. The performance of this scheme is independent of the polarities of symbols, and does not suffer from the "noise-cross-noise" effect. Simulations and comparison illustrate the well performance can be achieved with a small amount of data.