For resource allocation in IP over WDM networks, this paper proposes a novel routing and wavelength assignment algorithm based on layered-graph. The algorithm will open new wavelength paths if the arriving requests cannot be set-up over the existing IP-level topology effectively. As the developed integrated dynamic resource allocation policy combines the characteristics of the IP layer and optical layer, the algorithm naturally has a bettor network usage than those traditional ways taking into account the two layers separately. Simulation results show that the developed scheme performs very well in terms of performance metrics such as the hops of the lightpaths and the network blocking probability.

This paper proposes an adaptive transmission based dynamic packet reservation multiple access protocol for flexible and efficient support of multimedia services with different, quality of services under dynamically changing traffic and channel condition. In this scheme, all idle slots are classified as Free, Long Available and Short Available. Long and short messages can be transmitted in idle period of voice traffic. At the same time, protocol can adjust dynamically according to the number of Free slot the criterion of instantaneous traffic and adaptive transmission scheme is used to increase system efficiency further. Computer simulation confirms that the proposed system with perfectly selected parameters is effective in achieving larger capacity of multimedia communication systems.

Lots of research fruits have been presented in the modeling and improving methods of single TCP connection, while few in multiple TCP connections. This paper presents multiple TCP connections modeling through theoretical analysis under two cases: receivers with same characteristics and receivers with different characteristics, and obtains the RTT and long-term TCP throughput formulas related to number of reveivers, packet loss rate, and end-to-end delay. The simulation results confirm their effectiveness.

Wavelength continuity constraint leads to inefficient utilization of wavelength chan-nels and results in higher blocking probability in WDM optical transport networks. In order to weaken the inefficiency of the constraint, wavelength rerouting is used. Based on a wavelength graph model, the QoS-based wavelength rerouting strategy, dynamic selection method, is pro-posed. Simulation results of the network performance in different dynamic traffic loads are given. The results show that the method can not only meet the different QoS requirements of the client layer’s service, but also utilize the finite networking resources efficiently with reduced blocking probability.

A new unconstraint multicast routing algorithm based on genetic strategy, using Boltzmann acceptance rule and two populations, is presented in this paper. It can accelerate the speed of evolution, improve searching efficiency during the evolution process and decrease the probability of premature phenomena. It can get satisfactory multicast tree.

A linear approximation connection admission control scheme based on CLR of CBR sources in a finite buffer system and the lowest rate replacing is presented. Compared with CAC presented early, CAC in this paper has two advantages: the decision to accept or reject a new connection can be made more accurately by using a more rational and precise computation of CLR; QoS can be well guaranteed.

Ethernet is a widely deployed LAN technology. With carrier sensing and the corresponding collision backoff, traditional half duplex Ethernet with CSMA/CD does not need any other explicit flow control method. With the appearance of fast Ethernet and Ethernet switch, how to efficiently use and maintain network bandwidth and the resource of Ethernet switch in full duplex Ethernet is a key issue. In this paper, a queue length-based PAUSE algorithm, QL-PAUSE, is proposed which implements IEEE PAUSE operation based on input queue length. And through simulations, it is shown that QL-PAUSE can provide flow control and resource management in full duplex Ethernet, while achieving better performance and higher throughput.

After proposed the generic framework of the VPN service management, the shortage of the current management information modeling methods in the network/service management is analyzed and the advantage of the ODP/UML based modeling method is given. The applying ODP/UML for the management information model in the VPN service management is studied in detail. The implementation of the VPN SMS using CORBA is also given.

The 3G mobile communication system based on CDMA technique is expected to provide multi-services such as speech, video, email, FTP, Web browsing, etc. This paper presents a general analysis of the characteristics of the interference and capacity of the uplink of a multi-service cellular CDMA system. The system outage probability and capacity under different mixed traffic circumstances are analyzed respectively by approximating the interfer-ence with Gaussian and Lognormal distribution. The simulation results show the Lognormal approximation is more accurate than the Gaussian one. In addition, the impact of the shad-owing, power control error and imperfect sectorization on the interference and system capacity are evaluated.

In this paper, a new packet medium access control protocol, namely, Minislot Sig-naling Access based on Distributed Queues protocol is proposed in voice and data integration CDMA network (MSADQ/CDMA). Through proper manage of the PN codes, the number of random competition collisions reduces greatly, the Multiple Access Interference (MAI) de-creases, and the code resources are saved. Computer simulations are carried out in voice and data integration networks, and comparisons with other two reservation protocols are devel-oped, which show that MSADQ/CDMA protocol occupies less PN codes, but still has very good performance.

In this paper a novel joint CDMA/RTDMA MAC for multimedia transmission in wireless network is studied. In this scheme, the resource unit allocated by the base station is changed from a slot to a code-slot. Thus it enhances the system’s capacity and can support multi-media service more efficiently. The voice packet dropping probability and data delay are analyzed through M/M /n/m model and phase process respectively. Compared to the original RTDMA, this new MAC protocol can provide multiple services’ QoS even under the heavy load.

An MAC protocol named PMCAP/CDMA is introduced and utilized in voice/data integration network for uplink channel of CDMA mobile communication systems. Mathemati-cal model is built and performance analysis is conducted. Three code sets for voice contention, data contention and voice/data reservation are adopted so that there are no contentions be-tween voice and data users. Two adaptive code-set-assign methods are proposed. Simulation shows, compared with fixed assigned method, adaptive method can greatly improve the system performance. And the second adaptive code-set-assign method is a good trade off between voice drop rate and data delay.

In this paper, a new scheme based on digital certificate is proposed to implement secure multicast. In the scheme, tree-type multicast topology (multicast tree) is adopted. Each node of the multicast tree possesses a digital certificate which may indicate its identity in the multicast group. Also the certificate may be used for distributing session-key and authenticating session data. So the complexity of managing multicast group key is reduced. And because of tree-type structure of the multicast group, the joining and leaving for multicast member is more scalable.

The k-crror complexity of a periodic sequence of period N is denned as the smallest linear complexity that can be obtained by changing κ or fewer bits of the sequence per period. This correspondence shows the relationship between the linear complexity and the minimum value k for which the κ-error complexity is strictly less than the linear complexity, where period N equals to 2pq, q is an odd prime and 2 is a primitive root of mod q².

In this paper, a number-limited and time-limited proxy signature scheme is pro-posed based on elliptic curve. In the scheme, original signer can limit the number of proxy signature and recall the right of proxy signature as the valid time past. It overcomes the short-comings of universal proxy signature schemes that once original signer delegates his signature right to proxy signer, he can not limit the number of proxy signature and recall the right of proxy signature. This scheme that is constructed based on elliptic curve with one-time pair of keys has the merits of security, efficiency, smaller key and easy realization.

QoS of IP telephony is worse than that of circuit telephony because the available bandwidth of Internet varies then results in the loss of voice packets . This paper suggests an adaptive voice codec which can be applied in the IP telephony gateway, it can output various bit rate when Internet’s bandwidth varies, the most advantage of the codec is that it can decrease the ratio of packet-losing and improve the QoS of voice. In the implementation of the voice coder, this paper brings forward four algorithms, which include an algorithm for computing the ratio of packet-losing based on real-times transport protocol, an algorithm for implementing a coder that outputs various bit rate, an algorithm for voice packeting, and an adaptive algorithm for encoding and packeting.

In this paper, a joint source channel coding scheme used to transmit progressive image bit stream over noisy channel is proposed. The method can dynamically adjust the rates between source coding and channel coding, thus yields good performance and high efficiency. It still has the advantages of simplicity in system structure and ease in realization. Simulation results show that this scheme can highly improve the image quality compared to EEP and UEP scheme proposed previously especially when the channel is in bad condition.

This paper proposes a new adaptive EP latency change estimation algorithm based on the fractional lower order moment and the nonlinear transform for the error function. The new algorithm referred to as the NLST suppresses the lower order a-stable noises without the need to estimate dynamically the a values. The computer simulation and data analysis results show that this algorithm is robust under the lower order a-stable noise conditions, and it also provides a better performance than those of the DLMS, DLMP and SDA algorithms without the need to estimate the a value of the EP signals and noises.

This paper presents four data-selection schemes based on chip-matched filter for the multiuser receiver in multipath CDMA channels, and studies the impacts of these schemes on Multiple-Access Interference (MAI) and Inter-Symbol Interference (ISI). Then, it presents the Least Mean-Square (LMS) and Least-Square (LS) blind space-time multiuser detection methods both based on the Rosen’s gradient projection, and analyzes the effects of these schemes on the computational complexity of the two detection methods. Finally, a reasonable and realizable method is provided by analyzing the results of the simulations, and this method has the advantages of lowering the computational complexity, accelerating the speed of convergence and improving the detection performance, therefore it is interesting for application.

In this paper, it is verified that the eigen-vector corresponding to the maximum eigenvalue of the correlation matrix is arranged as an unsymmetrical matrix by definite rule (named eigen-matrix) and there are also signal-subspace and orthogonal subspace in the sigular value decomposition of the eigen-matrix. The orthogonal vector spectral estimation method based on the orthogonal subspace of eigen-matrix, deriving from signal subspace of correlation matrix, possesses high statistical stability, and it is an orthogonal method as well, so it is of high resolution. The eigen-matrix arranged by first eigen-vector processes lower dimensionality, so no pseudo peak appears. The new method is abbreviated as OVSS (Orthogonal Vector spectral estimation method based on correlation matrix Signal-Subspace). Lots of Monto-Carlo simulations have verified that the new spectral estimation method-OVSS is of high resolution, high statistical stability and less increments of calculation burden.

How to solve the computation complexity of the Dempster rule is a hot topic at present. In order to improve the accuracy of decision, this paper presents an approximation method which improves truncated D-S algorithm. Simulation results show that this method is superior to truncated D-S algorithm in terms of the accuracy of decision based on basic probability assignment function.

A simple method which for generalizing the classical orthonormal wavelets is pre-sented using investigation of the simplest Haar scaling function. A new kind of orthonormal wavelets is constructed from a classical orthonormal wavelets using the method proposed in this paper. These new wavelets inherit some basic properties of the corresponding classical wavelets, such as the orthonormality, the order of regularity, the time-frequency localization characteristics, and so on, meanwhile some performances of the new wavelets are improved. The generalized Haar wavelet, the generalized Shannon wavelet and the generalized Meyer wavelet, the generalized Daubechies wavelets are discussed emphatically. Finally, based on the new wavelet system, some fast algorithms for analytic wavelet transform analysis of real signals are studied.

A.method for analyzing variance-stationary random signal is presented. It is des-ignated as signal decomposition method. With the presented method, the trend component series can be obtained from the variance-stationary random signal; the rest of it is zero-mean stationary random signal that can be studied with the parametric model of stationary random signal. This method, simple and commonly used, is the general method to process the signal. The trend component series derived from this way can near more exactly the trend component curve of the variance-stationary random signal.

This paper proposes a multistate document segmentation method based on wavelet transform and the hidden Markov tree (HMT) model. With the method, each document is segmented into three textures, background, text and image. Furthermore, the method is improved by using the smooth image and a new segmentation method: IHMT segmentation is introduced. Finally, by examples it is illustrated that the methods are more effective than the two-state HMT model proposed by H.Choi et al.

A new method which performs classified vector quantization to wavelet image using space vector combination is presented. The method rearranges the wavelet coefficients utilizing fully the space and frequency correlation of them, and classifies the combined vector through joint judgement using the vector energy and zerotree vector. The whole image only need one quantization codebook, the classified information need a little bits to express, and vector quantization adopts weighted mean square error rule that using mankind visual characteristic to improve quantization gain. Simulation shows that this method can be realized easily and can compress the wavelet image efficiently under lower coding rate.

A new method which performs Trellis Coded Quantization(TCQ) of wavelet co-efficients based on multi-stage zerotree coding is proposed. First, the zerotree coding utilizes correlation between the subbands, then convolutional coding and signal space expanding are used for increasing Euclidian distance between signals. Finally, Viterbi algorithm is used to find a optimized survived quantized sequence.Simulation states that this method has an advantages of 0.3dD or so over optimized scalar quantization after zerotree coding. The method also has such capability that it has only modest encoding complexity with simple decoding.

A method called Two Dimension Trellis Coded Vector Quantization(2D-TCVQ) which applies the method of TCQ after VQ is implemented is proposed. This method first expands the small codebook to larger virtual eodebook, then uses Viterbi algorithm in two dimension to search the minimal distortion quantized route. The expansion of codebook min-ishes the minimal distortion in every subset and improves the quantization performance. By using small size codebook 2D-TCVQ can be applied in many low-power encode/decode con-dition. Simulation states the method has advantage of O.SdB nearly over TCVQ.The method also has capability that it lias only modest encoding complexity with simple decoding and it is insensitive to channel errors.

A new kind of median-based filtering algorithm-multi-window adaptive filtering is given to restore images corrupted by salt-pepper impulse noise. The size of the filtering windows is variable in the operation. Simulation results demonstrate that this algorithm is better than traditional median-based filters and particularly effective for the cases where the images are very highly corrupted.

The performance of an association algorithm for tracks generated by radar and IR, sensors in terms of the probability of correct and false association is analytically evaluated. The closed form of correct association probability versus the decision threshold and the polynomial formulation of the decision threshold versus the correct association probability are derived; the closed form of false association probability for a fixed value of correct association probability is obtained; and the closed formulation of correct association probability for prescribed false association probability is also given. The derived closed formula can be easily applied to the performance evaluation of radar and IR track association and the design of radar-and-IR data fusion system.

合成孔径雷达; 图像压缩; JPEG; 区域自适应

In this paper, based on the geometrical relationships in missile-borne Synthetic Aperture Radar(SAR), the model of radar signal and the method of imaging are discussed. In the imaging process, the effects of errors in measuring the maximum and minimum azimuth angles, and the missile’s velocities in three directions on positioning of missile-borne imaging arc analyzed in detail. The corresponding expressions are deduced, simulations in the state of high undershooting and nonlinear movement and experimental results in matching guidance are presented.

This paper describes the experimental study on repeat-pass airborne SAR interfer-ometry. The key technologies include airline control of aircraft, baseline determination, motion compensation and signal processing. In this paper a motion compensation method based on INS/GPS data has been proposed and implemented. An interpolation approach in range and azimuth dimensions is used to eliminate the effect of non-parallel baseline on correlation. Some resultant interferograms demonstrate the validity of the mterferometric L-SAR.

The dual-frequency and dual-polarization microstrip antenna for space-borne SAR application is presented. The antenna uses a single aperture, and folded configuration with aired plastic film. The patch antenna has low substrate dielectric constant and wide bandwidth. The cross-polarization isolation and impedance matching are discussed. The characteristics of the space-borne SARs using this light configuration antenna are analyzed.

The principle of multiple-port interferometer SAR/GMTI is analyzed and the in-fluence of various system errors on the capability for suppressing the main lobe clutter is also discussed in this paper, then the modification on the most critical weighting factor of this multiple-port interferometer is proposed. The theoretical analysis and simulation results show that this improved multiple-port interferometer SAR/GMTI scheme is insensitive to system errors or noise and is very robust, therefore it is suitable to be realized in the engineering.

A Generalized Forward-Backward Method (GFBM) with Spectral Accelerate Al-gorithm (SAA) to solve the Magnetic Field Integral Equation (MFIE) is developed. It is applied to numerical simulation of bistatic scattering from one-dimensional fractal dynamic sea surface with a ship presence under the TE tapered wave incidence at Low Grazing Angle (LGA). Nu-merical simulations of bistatic scattering at LGA show the functional dependence upon the fractal dimension of conducting rough surface, ship location and other parameters.

In this paper a synthesis for low power circuits is studied and an adiabatic ratioiess dynamic memory circuit is expressed quantitatively in accordance with the theory of three essential circuit elements. Then many adiabatic ratioiess dynamic flip-flops are composed of two adiabatic ratioiess dynamic latches, for example D or T’ flip-flop of 6 MOS transistors and D flip-flop with AND-NOR-inputs of 9 MOS transistors, in which there is no phenomenon of information disappearing rapidly after receiving one in capacitors. On the basis of above theory, this paper presents a synthesis for adiabatic synchronous sequential circuits, and designs an adiabatic 5421BCD decimal counter circuit of 32 MOS transistors which consumes lower power than that of an adiabatic PAL-2N 4-bit binary counter. Above theory is vertified by computer simulation.

The problem of 2-dimensional location using measurement provided by 3 sensors is minimum-determined problem. The solution of geometric method is obtained by direct substitution of the measured quantities and could not use the statistical information of the range measurements. In this paper, a new weighted geometrical location method is proposed for minimum-determined problem by transforming the range measurement between two points into the range measurement between a point and a line which is a known function of the actual range measurement. The new method obtains the source location by solving a weighted least square problem and the statistical information of the range measurements is used. Simulation studies show that the new location method outperforms the geometrical location method by reducing the root mean square error for scenarios with small range measurement errors.

Fair non-repudiation protocols resolve the problem of non-repudiation during the course of transferring information in electronic commerce and make all sides in equal positions. That is to say, the protocol is fair for each side. In this paper a fair multi-party non-repudiation protocol in "optimistic" sense is proposed. In the protocol, under normal conditions TTP is not needed. Only when either side is not honest or the communication collapses, is TTP involved in. This results in an efficient fair multi-party non-repudiation protocol.

With the development of SAR processing techniques, high image precision and high real time rate become an important index, especially on military field. This paper presents a medium granularity parallel processing algorithm for SAR imaging. In this parallel processing algorithm, every processing stage is done in parallel, and the degree of parallelism is task-level. It is fit for the parallel computer with good communication capacity. The experiments on DAWN1NG3000 show this parallel processing algorithm can get good results on real time rate and processing efficiency.

In this paper, by combining wavelet transform and Mahalanobis range, a feature extraction method of radar target using ISAR Imaging is presented. By processing ISAR imaging data using wavelet transform, giving a threshold, compressing data, and computing the Mahalanobis range of the data, the target feature series is obtained. The experimental result proves that the series is robustly invariant and can be used to identify radar target.

Several space-time code schemes available arc analyzed systematically from the basic prinpieal of space-time coding design. Some important problems in future research are proposed, and also the possible applications.