Wavefront curvature defocus effects can occur when Polar Formatting Algorithin(PFA)is used to reconstruct the SAR image. The traditional approach to this situation is simply to limit the size of the scene reconstructed. From the aspect of super-resolution, it is demonstrated in this paper that by accurate estimation and post compensation of the space-variant Quadratic Phase Errors(QPEs), super-resolution image and target features can still be obtained in the presence of wavefront curvature effects. The estimate. of QPEs is roughly obtained by use of the Discrete Polynomial-Phase TranSfornl(DPT), then it is refined by a RELAX-based parametric estimation rnet.hod. Experimental results show the effectiveness of the, proposed method.

The image processing of spaceborne Synthetic Aperture Radar (SAR) needs to process enormous amount of data and should accomplish the processing a.s soon as possible, hi order to increase the speed of SAR image processing, the paper presents the Chirp Scaling (CS) algorithm in parallel mode for the Dawn 2000-Ⅱwhich is a grand parallel computer. The paper also simulates the real time imaging system, which will work on satellite, and modifies the structure of the real time system according to the results of the simulation. The experiment proves that the parallel CS algorithm provided in the paper can get satisfying results on Dawn 2000-Ⅱ.

In the paper a new kind of reduced-dimension STAP (IV-STAP) is developed. At the same time its performance is compared with that of the 3B-4D-STAP, the 3DT-SAP and the full dimension STAP for there exist array errors or riot.

This paper discusses the problem of estimating the spatial signature of signals under the gains and phases of sensors unknown with a uniform linear array (ULA). It is utilized the redundant structure of the ULA and subspace-based ESPRIT-like algorithm is introduced to obtain, a closed form solution for the spatial signature of signals, and the gains and phases of sensors is presented. The major advantage of the proposed ESPRIT-like algorithm is its low computational burden compared with the other iterative methods and its robustness Compared with ESPRIT algorithm. Numerical simulation is presented to test the validity and the performance of this algorithm.

This paper, taking a view of engineering application and based on the asymp-lotics of signals, analyses practical algorithms and features and limitations in practice of three signal demodulation methods: Hilbert transform method, wavelet based multi-scales method, and asymptotic wavelet ridge extraction method. The advantage and practice problem of the asymptotic wavelet analysis approach, in obtaining the amplitude and frequency modulating information simultaneously, are illustrated with an authentic example of structure vibration analysis.

This paper proposes an adaptive early packet, discard scheme for TCP over ATM networks. Compared with general early packet discard scheme, this new scheme considerably rcduc.es the buffer size needed in the switch node for given QoS, improves the goodput and ininimizess the packet, loss.

A new grouped parallel multiuser detector is proposed and the corresponding algorithm is derived in this paper. The complexity of this detector is linear to the number of users and the performance is better than the decorrelating detector. The detection s delay is smaller than the decision-feedback detector. Numerical and simulation results show that the bit error rate (BER) of the grouped parallel multiuser detector approaches some value when the multiple access interference (MAI) power is 3dB higher than the useful signal power, so this detector can resist near-far effect efficiently for CDMA mobile communications.

It is suggested by Du Huiping(1999) to construct a novel multiple-beam adaptive antenna with Fresnel zone phase-correcting focusing configuration and feeding-array. In order to discover the operating mechanism of this new adaptive antenna, under a comparable uniform model, statistical characteristcs of feeding-array output, is analyzed by comparison between FMBA and ULA. Qualitative analysis shows that this type of adaptive antennas can realize adaptive signal processing effectively.

A micro-machined single crystal beam resonator for high accuracy pressure sensors by utilizing Micro-ElectroMechanical System (MEMS) technology has been successfully fabricated. By means of thermal excitation, the frequency response (including amplitude and phase) of the resonator has been characterized with an open-loop frequency sweeping method. The -3dB width of resonant peak (less than 1.2Hz) corresponds a very high quality factor Q value greater than 33000 in vacuum at room-temperature. The frequency response of the resonator at different levels of excitation has been discussed theoretically and experitrientally The result shows a good linearity of frequency versus excitation power at lower excitation power, and is coincident with the theory.

In this paper,the test property of fault subnetworks has been discussed based on the fault subnetwork equivalent current vector, and the essential condition for determinting fault subnetworks has been given. Using this method, every subnetwork may be diagnosed at the same time independently. The method not only may quicken diagnosis process, but also may solve the error transfer question. Thus, this paper should have theoretical and practical significance for tolerance fault diagnosis.

In this paper, the wavelet transform is used to increase the invisibility of the digital image watermark because it can exploit the temporal and spatial phenomena of human visual system. The more robustness can be achieved if watermarks are first spread and then modulated with pseudo noise spread sequence. The multiple address characteristic of spread spectrum will also help multiple watermarking. The performances of the proposed digital image watermarking algorithm based on wavelet transform combined with spread spectrum theory are discussed in this paper, such as resisting the damages from JPEG compression, random noising, geometrical transformation and multiple watermarking.

This article analyzes the different needs of RTOS in control system and signal processing system, and gets a conclusion that if reduce the requirement, a system software which gives attention to high efficiency and hardware independence simultaneously can be implemented.

A new hybrid genetic algorithm (HGA) that incorporates the flexible polyhedron method into the conventional genetic algorithm is proposed. The HGA speeds up convergence before the genetic algorithm reachs the global optima, and achieves better performance by performing global search and local search alternatively. The experiments for parameter estimation of nonlinear regression modelings show that the HGA has better generalization and effectiveness.

In this paper, a new method of synchronization using only software, which is applied to Digital Signal s Spectrum Coding (DSSC) for data transmission, is discussed. The step and the approach to achieve synchronization are presented. The advantage of this method is also discussed in comparison with traditional carrier synchronization.

In this paper, a method to improve the period property of (quasi-)chaotic digital filters by inserting a nonlinear transfer function is presented. Simulation results show that the resulted chaotic sequences have siginificantly increased periods while still preserving good correlation. Moreover, the structure is well suited to be realized in digital hardware and is thus attractive for applications to digital communications.

The difficulties that exist in classical electromagnetic field theory are discussed and it is pointed out that the Maxwell s equation set in classical theory is not self-consistent and only the approximate solutions can be gotten in classical electromagnetic theory. And how to establish the modern electromagnetic field theory based on modern mathematics and concepts of modern physics is given for solving this problem. By the modern electromagnetic theory, the mathematical form for macroscopic electromagnetic field theory is obtained and the exact solutions for electromagnetic field boundary value problem can be gotten.