摘要:In this paper, a segment-rebuilding method for active networks is presented. The principle is that by dividing the active network into a lumped element subnetwork and device subnetworks, and by means of the indefinite scattering matrixes of subnetworks and the connections and terminating of floating ports, the scattering matrix of the active network is founded for analysis and design of the network. The method can be used to deal with active circuits with complex topology structure. Some examples are presented to illustrate the appli-cations of this method.
摘要:A new channel routing model, while allows horizontal and vertical wire seg-ments to be placed on both layers, and two wires on different layer can be overlapped, is pre-sented. Based on this model, the constraints as well as the rules to layer assignment and routing order of nets in the two-layer channel routing are obtained. Then the property of opti-mal overlapping in a zone is analysed. By means of this property, the algorithm determines layer assignment and routing order of nets acoording to global optimat analysis. The routing area and number of vias are reduced much more than the router based on traditional Gv, Gh graph model. Preliminary experimental results are very encouraging. For the famous Deutsch’s difficult example, the algorithm obtains 18 tracks solution.
摘要:The silicon bipolar transistors with poly emitter suitable for 77k operation are presented in this paper. Temperature models of current gain at different current injection le-vels are given. The results indicate that current gain decrease more rapidly as temperature falls at lower current injection level. The trapping effect of the shallow-level dopants on cut-off frequency with various injection condition is discussed.
关键词:77K;Silicon bipolar transistors;Poly emitter;current gain;Cut-off frequency
摘要:Based on the piezoelectric model, this paper has studied the effect of piezoele-ctric charge on the depletion layer at channel/substrate interface of GaAs MESFET. It has been suggested that threshold voltage shifts of GaAs MESFET for positive piezoelectric char-ges are greater than those for negative. The interpretation of asmmetric inverse shifts of GaAs MESFET oriented in [011] and [011] directions is satisfactory.
摘要:Rate equations have been used for studying the dynamic response of a travelling-wave semiconductor laser amplifier (TWA) to a train of ultrashort optical pulses from an act-ively mode-locked semiconductor laser. The dynamic gain of the pulses acquired after passing the TWA and the carrier density distributions of the TWA have been found. The photon num-ber per pulse (corresponding to the pulse-energy) has been first used for studying TWAs.The dynamic gain saturation is determined only by the photon-per-pulse, regardless of input power. In addition, some issues concerned with the all-optical communication system have been dis-cussed.
摘要:In this paper, we analyze the effects of mode partition characteristics of semi-conductor laser diodes on optical fiber transmission system, discuss the physical meaning of mode partition coefficient K and calculate the transmission bit rate, distance and corresponding power penalty for the system in different bit-error-rate asymptotic line, fluctuation level, wave-length and optical source spectrum.
摘要:Based on the statistical distributions of the cepstral coefficients in the fea-ture space, this paper proposed a novel equivariance weighted cepstral distortion measure. Since the new weighting function reflects the fine structure of the feature space composed of speech cepstral vectors, the individual feature of speakers can be identified much efficiently. Preliminary experiments show that, for the VQ based speaker recognition, the proposed weighted cepstral distortion measure works substantially better than the Enclidean distance or the inverse variance weighted cepstral distance.
摘要:The neural net model, the feature extraction and the learning algorithm for the speech recognition by neural nets are investigated in this paper. A non-uniformly windowed pyramidical architecture is proposed, the normalizations of the feature vectors of two extraction algorithms are investgated and the Polak-Ribiere learning algorithm is modified to guarantee the weight vector not to converge to non-stable local minima. Speaker-dependent experments show that over 99% accuracy can be achieved for recognizing five vowels and ten digits. Speaker-independent experiments involved 90 speakers from 25 provinces of China show that 90.3% accuracy can be obtained for the digit recognition, and the performance can be increased by 1.0% after the pre-processing window is employed. The windowed model is more applicable for speech recognitions.
摘要:A new adaptive transform image coding, which is based on the analysis of features for image edges and brain mechanism of vision, is presented in this paper. The image sub-blocks are classified according to the control information of the image edges in spatial domain firstly, then resampling, quantizing and coding are realized according to different features of these classifications in transform domain, particularly some rectangle filters are de-signed suitable to data compression of the image sub-blocks which are part of the lines and directional edges in the image. Finally, the results of computer simulation with the example of a standard image of human head and shoulder are given. The experimental results show that the subjective quality of the coding image is good and the PSNR reaches 33.9dB when the code rate is as low as 0.2bit/pel.
摘要:Because most FORTRAN programs spend most of their time inside loops, ef-ficient exploitation of the loop parallelism is one of the most crucial activities during progra-m parallelization. Although loop parallelization is researched extensively and deeply existing techniques do not handle loops in a satisfactory manner. In this paper, we present a new te-chnique exploiting the irregular parallelism across loop iterations efficiently and completely.
摘要:Crystalline α-Si3N4 layer was formed by multiple N+ implantation into Si at high temperature, which omitted post high temperature annealing time. It is practically beneficail for simplifying working processes, lowering temperature and shortering time.
关键词:Crystalline α-Si3N4;High temperature implanta tion;Multiple N+ implantation
摘要:In situ formation of YBa2Cu3O7x superconducting film on (100) YSZ single crystalline substrates by long-pulse laser (150μs , l.06μm) beam shooting at the YBa2Cu3O7-x target has been demo-nstrated. The temperature of substrate heater is 750℃. During deposition, the oxygen pressure in vacuum chamber is about 6 Pa. The thin films prepared by this technique are shiny and strong, and its zeroresi-stance temperature is 84.7K. These films have been analyzed by x-ray diffractometry and scanning elect-ron microscopy.
摘要:An approach for CPM with noncoherent detection and 2-state trellis decoding is presented in this paper, and simulations are made for GMSK. It is shown that the bit error rate performance of GMSK, when decoded using noncoherent detection and 2-state trellis decoding, is far superior to the performance obtained using conventional one-bit differential detection.
关键词:CPM;GMSK;Noncoherent detection and soft decision
摘要:In this paper a novel design method for high-order all-poles OTA-C low-pass filters is presented. The 2nd-order to 7th-order Butterworth extremely flat low-pass filters have been designed by the method and simulatively analyzed with SPICE. Meanwhile it is proved that the design method pre-sented in this paper is strict theoretically and useful in engineering design.
摘要:The recursive-composition method is employed to solve the transfer functions of nonlinear systems by use of the recursive relation of Volterra responses and the transfer function formulae of com-position systems concerned.
摘要:The principle to construct hardware by which the fuzzy reasoning is ealized, is presen-ted in this paper. This hardware system can operate with analogous signal or discrite signal. It consists of far less fuzzy logical elements and has got high speed for real-time reasoning.
摘要:A convolutional algorithm of number theoretic transiorms (NTT) for computing a long linear convolution is proposed in this paper. By use of sborter NTT, this algorithm can calculate long convolution in the overlap-add algorithm. It reduces tbe number of bit-shift and the word length of convolutional machine, so that a reduction in the execution time of convolution is obtained.
摘要:In this paper, on basis that lesolving real coefficient polynomial is difficult, we present a kind of new system of public-key cryplography on binary probability and multivariale probabability systems and give the measures for eliminating calculation error.