A neural method to solve the orthogonality search problem arising in direction-of-arrival (DOA) estimation is proposed. The problem is mapped onto the snergy function of the neural network as a linear programming problem. The most important feature of this method hinges upon the fact that the number of neurons in the network is linearly dependent on the number of sensors in the sensor array, rather than the discretization levels of parameter space. Theoretical analysis and simulation results show that the p...

A loop-shift Stop & Go adaptive digital filter with applications in period detection and signal separation of overlapped periodic signals is presented. The filter places no contraints on the overlapped forms and requires no priori knowl edge except for approximate period range for signals to be separated, therefore it provides for an efficient and feasible method for separating periodic signals overlapped both in time and frequency domains. The convergence speed of the filter and the requirements of all parameter selections are discussed,meanwhile thecoarse precision period search and channel automatic close algorithm are proposed to reduce the amount of computation. Experiments show that the filter can separte all kinds of overlapped periodic signals with different initial phases and different relative magnitudes.

The Cross-spectral estimation methods are efficient in estimating the parameters of sinusoidal signals embedded in colored noise. But up to now there are only FFT and Cross-periodogram methods used in cross-spectral estimation. This paper introduces the modern auto-spectral estimation methods into cross-spetral estimation. The cross-correlation based Yule-Walker equation is proposed theoretically and the Moment and SVD algorithms for cross-spectral estimation are developed on the base. Finally a numerical example is given for comparing the presented methods with the well known Cadzow's SVD method.

A new calibration method for array channel mismatching is presented. A rough interval of a single calibration source difection is only known in the method. The method can be applied to arbitrary array geometries and suited to low SNR case. In addition, the method is rather robust for array channel mismatching errors. The computer simulations are given to support the proposed method.

This paper studies the problem applying Radial Basis Function Network (RBFN) which is trained by the traditional Recursive Least Square Algorithm (RLSA) to the recognition of one dimensional image of radar targets. The equivalence between RBFN and the estimation of Parzen window probabilistic density is proved, it is pointed out that the I/O functions in RBFN hidden units can be extended to general Parzen window probabilistic kernel function or potential function, too. This paper discusses the effects of the shape parameter α in RBFN and the forgotten factor λ, in RLSA on the results of the recognition of three kinds of kernel function such as Gaussian, Triangle, Double-exponential kernel functions, at the same time, and discusses also the relationship between λ and the training time in RBFN.

A new method that seek out best binary code by means of analytics is proposed in this paper, and it was proved that there is no n> 13 Barker code as well.

A simple and useful decision feedback equalizer used for non-linear channels with severe linear distortion and mild non-linear distortion is proposed. It is a combination of a non-linear channel equalizer based on connectionist model and a common decision feedback equalizer for linear channels. For a typical non-linear channel model it is shown by simulation that the equalization performances of the proposed equalizer are improved significantly.

A bursty traffic model is introduced in this paper to describe the statistical characteristics of packet video. The performance of leaky bucket algorithm with bursty traffic input is analyzed. The influences of various parameters on QOS (Quality of Service) are investigated. The analysis shows that although loss probability decreases through expanding buffer capacity, delay and delay jitter increase, whose effect on QOS will not be negligible.

A new approach to calculate fault detection probabilities is presented. The fault-free simulation is used in this approach. The quantities of controllabilities and observabilities are defined as probabilities of controlling and observing the nodes in the circuit. The probability distribution of random signals at circuit nodes is analyzed, and it is proved that controllability is in normal distribution. Thus the unbiasing estimate of fault detection probability can be obtained. According to the concept of observability, the fanout nodes are dealt with. For actual circuits the fault coverage obtained from this approach agrees favorably with the fault simulation results.

This paper deals with the design of switched-current (SI) high-pass ladder filters using a novel bilinear/feed-forward differentiator (BFD). The BFD is based on SI modular block, the generalized differentiator(GD) proposed in this paper. Various differential functions can be performed by the GD. Two different design techniques employing S domain and Z domain frequency transformations are described respectively. Examples of Chebyshev and Elliptic high-pass filters are also given.

Gram-schmidt algorithm is an interesting theme in the field of adaptive beamformer as a fast algorithm. However, a key problem associated with this algorithm is the detection of orthogonalisation on the basis of prior-knowledge of interferences dimension. In this paper, a simple and fast approach is proposed to detect the orthogonalisation automatically in the procedure of GS orthogonal decomposition. Finally, computer simulations are presented.

This paper presents a set of basic criteria to assess discretization schemes, which provides for choosing a proper discretization scheme of current continuity equation. According to the basic criteria, this paper also presents a new finite difference scheme. Both error analysis and numerical results show that the present scheme is superior to the Scharfetter-Gummel (SG) and Streamline Upwind Petrov/Garlerkin (SUPG) schemes in which the basic criteria are not satisfied.

The changes in the concentration and distribution of EL2 in undoped LEC SI GaAs caused by various annealing conditions were measured. The distribution of dislocation and precipitation of As were also measured for analyzing these changes. The origin of the inhomogeneity of EL2 distribution and the mechanism which improves the homogeneity of EL2 distribution by annealing are discussed on the basis of the experimental results.

The image reconstruction of a supporter is refered to an inverse problem of the defined domain of the integral which describes the physical process under investigation. A theorem is proposed. If the integrand satisfies some conditions, the contour of the domain can be reconstructed with this theorem. The results of computer simulation support the theorem and the analysis.

The Q factor of an electromagnetic open resonator due to the mirror reflection loss, mirror diffraction loss and coupling aperture loss has been analyzed in detail, and more precise formulae for calculating the Q factor are obtained. Finally a calculating example is given.

The resonant frequencies of the NBS TEM cell have been calculated by the Finite-Difference Time-Domain (FD-TD) and irregularly graded Ttansmis-sion-Line Matrix(TLM) methods. In the calculating process by the FD-TD method a novel processing method for the magnetic wall boundary conditions is developed with which it is possible to treat only one eighth of the whole cell domain to calculate the resonant frequencies. As a result, additional useful resonant frequency data are listed. In the TLM meth ,d, the prop...

This paper discussed a number of typical fast algorithms emerged in recent years for Code-Excited Linear Predictive (CELP) coding. They are classified into a number of basic methods, and their computational complexity and storage requirerments are compared.