Researches On Blind Estimation Of The Spreading Sequences For Direct-spreading Signals

The techniques of Direct Sequence-Spectrum Spreading(DS-SS), Code Division Multiple Access(CDMA), Multi-Rate DS/CDMA and so on are widely used in military and civil domain due to their predominance in anti-noise, anti-interference, low probability of intercept(LPI) and supporting mulit-rate data transmission. Investigating the non-cooperative signal processing technique for the systems, which are enumerated above, is of great importance. And the blinding estimation of the spreading sequences is topic currently. This dissertation focuses on the blinding estimation of the spreading sequences of signals, which are enumerated above. The main research works are listed as follows.1. For the joint blinding estimation of the spectrum spread and information sequences of the SC-DS-SS signal. A method, which based on 2-norm principle of vector space, is proposed to unbiased estimate desynchronizing time firstly. Then an algorithm, based on unbiased desynchronizing time, is presented to estimate the spectrum spread and information sequences jointly. The received signal is divided into two-spread-period-length temporal vectors overlapped by one-spread-period, and accumulates these vectors one by one to form the received signal matrix. And then, exploits operation of singular value decomposition(SVD) to the matrix. Then form vector space with the dimension equal to spreading period in the right singular vector, which correspondence to the max singular value sequentially. The position of initial point correspondence to 2-norm maximum is the unbiased estimation of desynchronizing time. And then intercept the spreading sequence according to desynchronizing time and spreading period. And the information sequence is then recovered blindly from the left singular vector. The method avoids the phase ambiguity in traditional algorithms which use two vectors to reconstruct the spreading sequence, and unlimited by the type of spreading sequences2. Blinding estimation of spreading sequence for long-code DS-SS signal is studied. A improved algorithm, which based on second order blind identification(SOBI) and multi-user model building, is proposed to estimate the spectrum spread sequences of the periodic long-code DS-SS(PLC DS-SS) signal. The method makes use of joint diagonalization and special constraint condition(such as m- or Gold-sequence) to eliminate the phase ambiguity, to form the blind estimation of long-code spreading sequence lastly. For the non-periodic long-code DS-SS(NPLC DS-SS) signal, the near-optimization segmentation criterion is extracted, according to spreading periodic and spreading gain. The deficiency of the OSEA algorithm, which depending experience to segmentation, is settled. The performance of the OSEA algorithm is optimized. In addition, the algorithm based on SOBI is advanced to blind estimate the special spreading sequence, such as m sequence and so on.3. The dissertation then addresses the problem of blind estimation of the spreading sequences of DS-CDMA signals. For short-code DS-CDMA signals, a data matrix is reformulated in a similar way as the short-code DS-SS signals first, and the desynchronization time is estimated by checking the maximum of the determinant of the data matrix. Based on the estimate of the desynchronization time, the Fast-ICA algorithm is introduced in to propose a method for the blind estimation of the spreading and information sequences. In this method, the spreading sequences of the users are extracted by exploiting the desynchronization time estimate, and then distinguish the information sequences of different users according to the similarity criterion in the estimated subspace. For long-code DS-CDMA signals, a more rigorous overlapping segmentation criterion is presented to substitute the empirical one, and the Fast-ICA algorithm is then used to estimate the local spreading sequences of each user respectively, and the local sequence estimates are finally synthesized to obtain the whole spreading sequences by removing the ambiguities in order and phase with respect to the correlation of the inter-segmentation overlapped sequences.4. The problem of blind spreading sequence estimation of the multi-rate direct-sequence code division multiple access(MR DS-CDMA) signals is also addressed in this dissertation, and two methods that are based on the Givens rotation and the Fast-ICA algorithm are proposed. The method based on the Givens rotation segments the received signals with which form by vectors with two-biggest-spreading-period-length temporal width, and then make use of Givens rotation to estimate the spreading sequences which with the biggest period blindly, make use of 2-norm to estimate the timing offsets. Then estimate the sequences which with small period via a matrix-deflation, noise-counteract manner iteratively. The Fast-ICA algorithm is used to estimate the spreading sequences blindly which with the same spreading period. The estimate PN sequences is obtained according to the desychronous time, which is calculated based on the 2 norm of vector, exploit the normalization weigh function of estimated waveform similarity ratio to separate spreading sequence which with different spreading period, from biggest priod to smallest period.