A new digital modulation recognition method is proposed which is based on the features extracted from generalized autoregressive (GAR) model parameters and the use of neural network classifier. Because of the better noise suppression capacity of the GAR model and the powerful pattern classification capacity of the neural network classifier, the new method can significantly improve the recognition performance in lower SNR. Computer simulations also show that the new method possesses higher recognition ratio and better robustness.

In this paper a frame-based packet reservation multi-access method is presented for data transmission in CDMA cellular system. Assumed that the number of arriving data packets follows a Poisson distribution, this paper analyzes the throughput and mean delay of data services using equilibrium point analysis (EPA), Simulation results indicate that the proposed scheme can support more data terminals than the traditional cellular system and achieve acceptable mean delay.

Recent measurements show that the traditional traffic models cannot be success on kinds of long-range dependent traffic which has great influences on the network performance. Fractal or self-similar models are found to be more suitable for describing the modern traffic. But there is still little knowledge about the performance of the multiplexer under self-similar traffic. This paper proposes a Quasi Self-Similar Process (QSSP). Using this process, the upper bond of the cell loss rate and multiplexing gain of the multiplexer in N i.i.d. QSSP input case is gotten. If the sources with different parameters, an efficient numerical algorithm to get a bond is also proposed. Simulation indicates that the analysis fit with the simulation result well.

A new approach for image restoration based on Cellular Neural Network(CNN) is proposed. The physical meaning of Maximum Entropy (ME) is analyzed and a new template is proposed for ME binary image restoration. The result of computer simulation proves this approach is reasonable.

An Expectation-Maximization(EM) training algorithm for estimating the parameters of a special Probability Mapping Network (PMN) structure which forms a multicatolog Bayes classifier is proposed in this paper. The structure of PMN is a four-layer Feedforward Neural Networks(FNN), where the Gaussian probability density function is realized as an internal node. In this way, the EM algorithm is extended to deal with supervised learning of a multicatolog of the neural network Gaussian classifier. The computational efficiency and the numerical stability of the training algorithm benefit from the well-established EM framework. The effectiveness of the proposed network architecture and its EM training algorithm are assessed by conducting two experiments.

In this paper, the fully-connected higher-order neuron and sparsed higher-order neuron are introduced, the mapping capabilities of the fully-connected higher-order neural networks are investigated, and that arbitrary Boolean function defined from {0,1}^N can be realized by fully-connected higher-order neural networks is proved. Based on this, in order to simplify the networks architecture, a pruning algorithm for eliminating the redundant connection weights is also proposed, which can be applied to the implementation of sparsed higher-order neural classifier. The simulated results show the effectiveness of the algorithm.

After briefly interoducing the characteristics of 1/f noise in millimeter wave focal plane array detectors, the paper analyses the relation of the wavelet transform and 1/f noise in detail and suggests the fashion of decorrelating 1/f noise using the wavelet transform., also deduces the relative expressions. The results of computer simulation show the good effect.

This paper consider blind equalization of QAM systems, a new blind equalization criteria based on second and fourth order moments is presented, and a new blind equalization algorithm is derived, it also improved the understanding of blind equalization problem in theory.

This paper proposes a multipart key escrow scheme, based on ElGamal public key cryptosystem and Simmons multipart secret-sharing protocol, which not only effectually prevents subliminal channels attack in "fair public key cryptosystem", but also suits more complicated escrow environment.

Double-loop network has been one of the most populular topologies used in the distribution systwms. An important problem is to find the shortest path between any two nodes in this kind of networks. This paper gives an O(△) time algorithm for finding the shortest path between any two nodes of the double-loop network.

In this paper, the vectors set called complete set is presented. A class of second order ergodic matrix is constructed by using the complete set, and a class of third order ergodic matrix is also constructed by using the partial set of complete set. On the basis mentioned above, the corresponding Boolean functions with second order and with third order correlation immunities are given. The paper also proves that the additional column vector is not allowed for both second order and third order ergodic matrices so constructed.

Based on the scale function representation for a function in L²(R), a new wavelet transform based adaptive system identification scheme used for acoustic echo cancellation is proposed. It can reduce the amount of computation by exploiting the decimation and keep the advantage of quasi-othogonal transform with the discrete wavelet transform(DWT). The issue has been supported by computer simulations.

This paper presents a minimum error thresholding algorithm under the hypothesis that the gray level histogram of SAR image fitting to a mixture model of shifted Rayleigh distribution. This algorithm is applied to real SAR images and compared with traditional Otsu algorithm and other minimum error thresholding algorithms based on various models of histogram. The hypothesis of Rayleigh distribution model is confirmed by Kolmogorov-Smirnov testing and the results obtained show that the proposed new algorithm has good performance in image thresholding for SAR images.

A new type of genetic algorithm combined the parallel simulated annealing with the advantages of guaranteeing convergence and deciding convergence easily was proposed to the synthesis of unequally spaced arrays. The position and the weighted coefficients can optimize at the same time. The cases showed that this method has strong ability to find the global optimization solution. This method provides a strong tool to the design of the large antenna arrays.

A new offset elliptic Presnel Zone Plate Antenna (FZPA) has been suggested based on the Fresnel principle. Simulation of the focusing property of this FZPA and that of the regular one showns that the one suggested here can give better focusing characteristics in some applications. The numerical results give a help to the design of offset FZPA.

When a dipole source(electric dipole or magnetic dipole) is located at arbitrary point in circularly multilayered conductive medium,the electromagnetic fields excited by the source contain both TE and TM waves.The series solutions of the fields in very layer are derived analytically in this paper.A coefficient recurrence matrix is established by means of the boundary conditions.In such way,the expressions of fields can be degermined definitely.As a numerical example,a group of response curves of an off-axis source are given finally.

A novel numerical modeling technique, based on FD-TD and the intrinsic characteristic of spatial periodic structure, is used to analyze the high frequency characteristics of the slow wave structure of arbitrary geometry. The method was sucessfully applied to calculate the complete dispersion relation and couple impedance of TM_0n modes in a coaxial corrugated wall waveguide.

The space-wave method, suitable for extended surface targets, is a preferable method for permittivity measurement of ground materials at microwave frequencies. The inversion of permittivity from the measured data, which is established on using the Fresnel reflection coefficients, is easy if the measurement includes phase information. However, if the measured quantities are just the amplitudes of the reflected waves, the inversion will be much complicated by solving a quadratic nonlinear equation set, and so far only numerical iterative approaches have been reported in the literature. Unfortunately, the iterative process is highly unstable because of the extremely sensitivity of the inverse algorithms to the measured data and even to the iterative initial values. In the present work, analytical inverse formulations are presented which will give a unique inverse result if it does exist for a set of measured data. Error sensitivity of the inversion to the measurement errors are performed, as well.

Taking account of the complexity, delay and realistic, this paper has discussed the realization arragement of the trellis-coded modulaiton applied to the multi-carrier modulation system. The simulation analysis has been taken in the ADSL DMT environments. According to the simulation results, the trellis-coded modulaiton can also obtain nearly the theoretical gain when used in the multi-carrier modulation system.

This paper presents the analysis of a continuous-time model of the asymmetric polling system with exhaustive service. Compared with the discrete-time model of the polling system with exhaustive service(by Zhao Dongfeng, 1994), the results of two polling systems are given.

In this paper, generalized Hamming weights for Goppa codes and BCH codes are studied. Lower bounds of generalized Hamming weights for Goppa codes are obtained and an algorithm to find the lower bounds is given. Moreover, the last few generalized Hamming weights for narrow and primitive BCH codes are determined.

The stability ot correlation-immune functions is discussed in tins paper.The correlation-immunity is not stable and generalized Correlation-immunity is stable.

This paper reviews the current status of High Definition Television (HDTV) briefly and sets forth brilliant application perspective of non-broadcasting HDTV. The design and developement of high definition image processing sub-system, and relative software support are mainly introduced. The sub-system has been used to "8.5" key project-non-broadcasting high definition color closed television system, and its performance meets to the requirements for the system. It is believed that the sub-system will be widely applied to military, meteorological remote sensing, movie and television, medical and hygiene, industry, printing and scientific research.

A key problem of permutation filters is its too many parameters. In this paper, modified permutation filters called self-dual permutation filters are proposed. The new class filters need half of the permutation filters parameters, but they give almost equal performances with permutation filters experimentally.

By applying symmetric transmission current-switch theory, the symmetric ternary current-mode CMOS Schmitt inverter which can have multiple outputs is proposed, and its width of hystersis can be adjusted as needed. The result of computer simulation shows that the designed Schmitt inverter has nearly ideal Schmitt function.