A key issue in spoken language processing has become the integration of speech understanding and natural language processing(NLP). This paper presents a parallel computational model for the integration of speech and NLP. The model adopts a hierarchically-structured knowledge base and memory-based parsing techniques. Processing is carried out by passing multiple markers in parallel through the knowledge base. Speech-specific problems such as insertion, deletion, and substitution have been analyzed and their parallel solutions are provided. The complete system has been implemented on the Semantic Network Array Processor(SNAP) and is operational. Results show an 80% sentence recognition rate for the Air Traffic Control domain. Moreover, a 15-fold speed-up can be obtained over an identical sequential implementation with an increasing speed advantage as the size of the knowledge base grows.

In this paper, we have analyzed the error performance of the optimum threshold detection(OTD) of 16 QAM signal in the Rician fading channel with and without the maximal ratio combining(MRC) diversity technique in the presence of cochannel Rayleigh interference. An also the error performance of OTD is compared to that of conventional threshold detection(CTD) in the Rician fading channel in the presence of cochannel Rayleigh interference. With the result of analysis, it is found that there exists a synergistic effect due to both MRC diversity and optimum threshold detection in the Rician fading channel in the presence of cochannel Rayleigh interference.

Adaptive code rate change schemes in DS-SSMA systems are proposed. In the proposed schemes, the error correcting code rate is changed according to the channel states. Two channel states having significant effects on the bit error probability are considered: one is the effective number of users, and the other is the fading environment. These channel states are estimated based on retransmission requests. The criterion for the change of the code rate is to maximize the throughput under given error bound.

In this series of two papers, adaptive code rate change schemes in DS-SSMA systems are proposed. In the proposed schemes the error correcting code rate is changed according to the channel states. Two channel states having significant effects on the bit error probability are considered: one is the effective number of users considered in Part 1, and the other is the fading environment considered in Part 2. These channel states are estimated based on retransmission requests. The criterion for the change of the code rate is to maximize the throughput under given error bound. Simulation results show that we can transmit maximum amount of information if we change the code rate based on the channel states.

To improve the reliability of control systems, certain robustness to plant uncertainties and disturbance inputs is required in terms of well founded mathematical basis. Robust control theory was set up and developed until now from this motivation. In this field, H$_2$or H\ulcorner norm performance measures are frequently used nowadays. Moreover a mixed H$_2$/H\ulcorner control problem is introduced to combine the merits of each measure since H$_2$control usually makes more sense for performance while H\ulcorner control is better for robustness to plant perturbations. However only some partial analytic solutions are developed to this problem under certain special cases at this time. In this paper, the mixed H$_2$/H\ulcorner control problem is considered. The analytic(or semi-analytic) solutions of (sub)optimal mixed H$_2$/H\ulcorner state-feedback controller are derived for the scalar plant case and the multivariable plant case, respectively. An illustrative example is given to compare the proposed analytic solution with the existing numerical one.

In this paper, adaptive controller with switching action is designed for rigid body robot manipulators to ensure the uniform stability of the manipulator system without a priori knowledge of the unmodeled dynamics. It will be shown that the parameter estimates are bounded independent of the other closed-loop signals boundedness, and also shown that the tracking error belongs to the normalized error bound via mathematical analisys. The robustness and performance of the proposed adaptive controller is investigated for the two-link direct drive manipulator actuated by VRM(Variable Reluctance Motor).

This paper mainly describes the wind load assumption of 765kV transmission towers. We analyzed wind velocity data a meteorological observatories to get the wind velocity of 50 years return period by using Gumbel I type extreme value distribution. By multi-correlative regression analysis method, wind velocity at no observation site was obtained. Reference dynamics wind pressure map was obtained from above analysis and the wind pressure was classified as three regio in high temperature season.

An improved predictive current control technique for a zero current switched(ZCS) buck-boost series resonant inverter(SRI) is proposed to overcome the inherent disadvantages such as the uncontrollable large overshoot and the large current ripple. Using the proposed technique, four quadrant operations of the output voltage and current for an uninterrutible power supply(URS) application are guaranteed and the buck-boost operation can also be obtained without an additional bidirectional switch.

This paper presents a new approach to derive an energy integral based on an Equivalent Mechanical Model(EMM), which is developed by introducing imaginary springs for line resistances. The proposed EMM shows that phasor currents and voltages are directly analogous to the two-dimensional force and displacement vectors, respectively. Through rigorous energy analysis of the proposed EMM, an exact energy integral expression is derived for multimachine systems, and several useful theorems are developed to derive an energy integral for power systems with detailed generator models the energy integral exactly reflects the internal resistance, saliency and flux-decaying effects of the generator. Finally, an illustrative example is given for a multimachine system adopting the Eq'-model for generators, which shows that the consideration of a detailed generator model does not aggravate the complicacy of the direct method of stability analysis in multimachine systems.

In this paper, we have proposed a band stop filter (BSF) for pre-processing of image sequences before encoding. By pixel-wise temporal filtering of the image sequences using the BSF, the bandwidth and noise of the signal are reduced, while preserving the image quality in view of human visual perceptions. As a result, when compared to the original image sequences, te pre-filtered image sequence requires lower bit-rates for encoding, while there is not much degradation in quality. Also, it has been shown that the proposed BSF causes less smearing and blurring than the conventional recursive low pass filter for bandwidth and noise reductions.

Considering the human visual masking property, a modified relationship between the activity function and the visibility threshold is developed. This leads to a novel objective appraisal method for blocking effects in a lossy subimage coding by virtue of the human visual sensitivity. The appraisal criterion is examined using a series of reconstructed images that are DCT-coded at various bit rates. Experimental results show that the presented blocking effect measure well agrees with the subjective ranking.

In this paper, we present an analysis of the IS-95 authentication protocol for the Code Division Multiple Access(CDMA) mobile communication network. We propose a mutual authentication protocol, AP-6, to improve the security and correctness. Formal description and verification of the proposed AP-6 are also presented on the basis of the formal logic. It is shown that the proposed AP-6 is more secure and correct than the IS-95 authentication protocol.

This paper proves by straightforward computation an interesting property of a recursive indexing: it preserves the entropy of a geometrically-distributes stationary memoryless source. This result is a pleasant surprise because the recursive indexing though one-to-one, is a symbol-to-string mapping and the entropy is measured in terms of the source symbols. This preservation of the entropy implies that the minimum average number of bits needed to represent a geometric memoryless source by the recursive indexing followed by a good binary encoder of a finite imput alphabet remains the same as that by a good encoder of an infinite input alphabet. Therefore, the recursive indexing theoretically keeps coding optimality intact. For this reason recursive indexing can provide an interface for a binary code with a finite code book that performs reasonably well for a source with an infinite alphabet.

In this paper, we propose a scheme that maps the time-delay neural network (TDNN) into the neurocomputer called EMIND-II which has the wavefront toroidal mesh-array structure. This neurocomputer is scalable, consists of many timeshared virtual neurons, is equipped with programmable on-chip learning, and is versatile for building many types of neural networks. Also we define the programming model of this array and derive the parallel algorithms about TDNN for the proposed neurocomputer EMIND-II. In addition, the computational complexities for the parallel and serial algorithms are compared. Finally, we introduce an application of this neurocomputer to word recognition.

In this paper, we propose a new type of recurrent neural network architecture in which each output unit is connected with itself and fully-connected with other output units and all hidden units. The proposed recurrent network differs from Jordan's and Elman's recurrent networks in view of functions and architectures because it was originally extended from the multilayer feedforward neural network for improving the discrimination and generalization power. We also prove the convergence property of learning algorithm of the proposed recurrent neural network and analyze the performance of the proposed recurrent neural network by performing recognition experiments with the totally unconstrained handwritten numeral database of Concordia University of Canada. Experimental results confirmed that the proposed recurrent neural network improves the discrimination and generalization power in recognizing spatial patterns.

In safety critical hard real-time systems, a timing fault may yield catastrophic results. In order to eliminate the timing faults from the fast responsive real-time control systems, it is necessary to schedule a code based on high precision timing analysis. Further, the schedulability enhancement by having multiple processors is of wide spread interest. However, although an instruction level parallel processing is quite effective to improve the schedulability of such a system, none of the real-time applications employ instruction level parallel scheduling techniques because most of the real-time scheduling models have not been designed for fine-grain execution. In this paper, we present a timing constraint model specifying high precision timing constraints, and a practical approach for constructing static schedules for a VLIW execution model. The new model and analysis can guarantee timing accuracy to within a single machine clock cycle.

The processor allocation problem in mesh multicamputers is to recognize and locate a free submesh that can accommodate a request for a submesh of a specified size. An efficient submesh allocation strategy is required for achieving high performance on mesh multicomputers. In this paper, we propose a new best-fit submesh allocation strategy for mesh multicomputers. The proposed strategy maintains and uses a free submesh list to get global information for free submeshes. For an allocation request the proposed strategy tries to allocate a best-fit submesh which causes the least amount of potential processor fragmentation so as to preserve the large free submeshes as many as possible for later requests. For this purpose, we introduce a novel function for quantifying the degree of potential fragmentation of submeshes. The proposed strategy has the complete submesh recognition capability. Extensive simulation is carried out t compare the proposed strategy with the previous strategies and experimental results indicate that the proposed strategy exhibits the best performance along with about 10% to 30% average improvement over the best previous strategy.

Loops are some of the richest program constructs where parallelism is available. Exploiting fine-grain parallelizm out these constructs is particularly important in light of the growing popularity of superscalar and VLIW machines. This paper explains how the fine-grain parallelization techniques can be generalized to handle nested loops. Our technique integrates nested loop parallelization techniques at the fine-grain level, thus exposing more fine-grain parallelism, and is flexible enough to handle non-perfectly nested loops. Examples and some experimental results are presented to illustrate our approach.

This paper derives the two-dimensional probability distribution and density functions of morphological dilation and erosion of a one-dimensional memoryless source and reports numerical results for a uniform source, thus providing methodology for joint distributions for other morphological operations. The joint density functions expressed in closed forms contain the Dirac delta functions due to the joint discontinuity within the dilation and erosion. They also exhibit symmetry between these two morphological density functions of dilated and/or eroded sources, in the computation of other higher moments thereof, and in multidimensional quantization.

Element failure occurs with high probability for every array used in the real world ; that is, it is a common phenomenon that there are one or more elements with no output. Element failure produces an elevated sidelobe level and fails to reject the interference signals in an adaptive beamformer. In this paper, we present the adaptive beamforming algorithm for array with element failure. The presented method minimizes the array output power subject to a set of linear constraints which maintain the frequency response in the look direction and force the weights of the inoperative elements to zero. Numerical results have been included.

New formulas for McClellan transform parameters for the design of 2-D zero-phase FIR fan filters are optimally derived under the integral squared error(ISE) criterion. By imposing the constraint that F(0, 0)=\ulcorner, where F($.$) is the McClellan transform and $\omega$\ulcorner is the cutoff frequency of the 1-D prototype filter, the ISE is directly minimized without modifying it and, as a consequence, closed-form formulas for the McClellan transform parameters are obtained. It is shown that these formulas lead to a very efficient design for 2-D zero-phase FIR fan filters.

The partitioning of related objects should be performed before clustering for an efficient access in object-oriented databases. In this paper, a horizontal partition of related objects in object-oriented databases is presented. All subclass nodes in a class inheritance hierarchy of a schema graph are shrunk to a class node in the graph that is called condensed schema graph because the aggregation hierarchy has more influence on the partition than the class inheritance hierarchy. A set function and an accessibility function are defined to find a maximal subset of related objects among the set of objects in a class. A set function maps a subset of the domain class objects to a subset of the range class objects. An accessibility function maps a subset of the objects of a class into a subset of the objects of the same class through a composition of set functions. The algorithm derived in this paper is to find the related objects of a condensed schema graph using accessibility functions and set functions. The existence of a maximal subset of the related objects in a class is proved to show the validity of the partition algorithm using the accessibility function.