Date 78 March 1995
Filter Results

Eighteenth Convention of Electrical and Electronics Engineers in Israel

Authors index

Recent results in universal coding for probabilistic sources and individual sequences
Summarizes some recent work in the field of universal data compression, both from a probabilistic and a deterministic point of view. In the probabilistic setting, the data x to be compressed is generated by a source from a certain class {P(x/spl theta/),/spl theta//spl isin//spl Lambda/}. The capacity C of the channel from /spl theta/ to a is well known to be the minimum attainable redundancy of universal lossless codes w.r.t. this class, both in the minimax sense and in the Bayesian (maximin) sense. We show that C is essentially a lower bound also in a stronger sense, that is, for "most" sources in the class. This extends Rissanen's lower bound for parametric families. In the deterministic setting, an individual sequence is assumed without any statistical assumptions, and we seek a universal lossless encoder that is essentially no worse than the best encoder in a certain class of encoders, e.g., finitestate (FS) encoders, block encoders, etc. We characterize the the minimum attainable redundancy of universal coding w.r.t the best encoder in this class. For the class of FS encoders, we derive a bound that holds for "most" sequences in a certain sense, which is analogous to the classical probabilistic minimum description length (MDL) principle. The bound is attainable sequentially by a suitable arithmetic code, and therefore a deterministic significance of the predictive MDL concept is established. Finally, we extend this approach to lossy compression and obtain results in the same spirit for blocktoblock and blocktovariable length codes. View full abstract»

On achievable information rates over slowlyfading channels with ideal and partial side information
Page(s): 1.1.1/1  1.1.1/3Slowlyfading channels with ideal or partial side information on the fading levels are addressed. We focus on informationtheoretic considerations, regarding two main issues: the effect of decoding delay constraints, and the effect of the amount of side information available to the receiver. Several channel models are discussed, and both new and known results are presented in a unified framework. View full abstract»

Encoding mismatch in lossy source compression
Page(s): 1.1.2/1  1.1.2/5A source sequence is described by the index of the codeword in some codebook C that is closest to it according to the distortion measure d/sub 0/(x,x/spl circ//sub 1/). We study the distortion associated with reconstructing the source sequence so as to minimize the distortion measured by d/sub 1/(x,x/spl circ//sub 1/), which is in general different from d/sub 0/(x,x/spl circ//sub 0/). Using a random coding argument we derive an upper bound on the resulting distortion. Applying this bound to blocks of source symbols we construct a sequence of bounds which are shown to converge to the least distortion achievable in this setup. This solves the rate distortion dual of an open problem related to the mismatched capacity of channels. Addressing a different kind of mismatch, we also study meansquarederror description of nonGaussian sources with Gaussian codebooks. It is shown that the use of a Gaussian codebook to compress any ergodic source results in an average distortion which depends on the source only via its second moment. The hardest source to describe is the memoryless Gaussian source, and it is best described using a Gaussian codebook. Once a Gaussian codebook is used, we show that all sources of a given second moment become equally hard to describe. View full abstract»

A technique to prove the channel coding theorem for a reduced class of codes and its applications
Page(s): 1.1.3/1  1.1.3/5The channel coding theorem for DMCs is proved using random coding. The standard proofs use the property of total randomness of the ensemble of codes. For a more restricted ensemble of codes a specific proof is needed. This has been done, for example, for linear codes, timevarying convolutional codes and for constant composition (constant type) codes. In this work we show that weaker demands on the ensembles of codes, than independency of the code words, are needed. As a result, we get simple proofs of the channel coding theorem for interesting restricted ensembles of codes. As an example we can prove the channel coding theorem for the ensembles of codes based on the output of an LFSR (linear feedback shift register). A particularly interesting application of our technique is a proof that fixed (timeinvariant) convolutional codes attain capacity. View full abstract»

A converse theorem for the multiple description problem
Page(s): 1.1.4/1  1.1.4/4Consider a sequence of discrete i.i.d random variables x/sub 1/,...,x/sub n/ and a distortion measure d(x,x/spl circ/) on the estimate x/spl circ/ of x. Two descriptions i(x)/spl isin/{1,2,...,2/sup nR1/} and j(r)/spl isin/{1,2,...,2/sup nR2/} are given of the sequences x=(x/sub 1/,...,x/sub n/). From these two descriptions, three estimates x/spl circ//sub 1/(i),x/spl circ//sub 2/(j) and x/spl circ//sub 0/(i,j) are formed, with resulting expected distortions: E{1/n/spl Sigma//sub k=1//sup n/d(x/sub k/,x/spl circ//sub mk/)}/spl les//spl Delta//sub m/, m=0,1,2. We find that the set (R/sub 1/,R/sub 2/,/spl Delta//sub 0/,/spl Delta//sub 1/,/spl Delta//sub 2/) is achievable only if there exists a probability mass distribution p(x)p(x/spl circ//sub 0/,x/spl circ//sub 1/,x/spl circ//sub 2/x) with E1/n/spl Sigma//sub k=1//sup n/d(x/sub k/,x/spl circ//sub mk/)/spl les/m=0,1,2 such that: R/sub 1//spl ges/I(x;x/spl circ//sub 1/); R/sub 2//spl ges/I(x;x/spl circ//sub 2/); R/sub 1/+R/sub 2//spl ges/I(x;x/spl circ//sub 0/x/spl circ//sub 1/,x/spl circ//sub 2/)+I(x;x/spl circ//sub 1/)+I(x;x/spl circ//sub 2/); where I(/spl middot/) denotes the Shannon mutual information. I(x;x/spl circ//sub 1/) and I(x;x/spl circ//sub 2/) are the necessary rates for achieving the expected distortions /spl Delta//sub 1/ and /spl Delta//sub 2/ respectively. The excess rate I(x;x/spl circ//sub 0/x/spl circ//sub 1/,x/spl circ//sub 2/) should be added to I(x;x/spl circ//sub 1/) and I(x;x/spl circ//sub 2/) for ensuring that x/spl circ//sub 0/ has distortion less or equal to /spl Delta//sub 0/. This result has an interesting intuitive interpretation. View full abstract»

Parametric estimation of single morphologically varying evoked brain potentials
Page(s): 1.2.1/1  1.2.1/5We present a new parametric estimator for single Morphologically Varying Evoked Potentials (MVEP's). The estimator can detect changes in specific components within the Evoked Potential (EP) complex, extending current methods which only compensate for global latency and magnitude variations with respect to an averaged evoked response. The performance of the estimator is analyzed analytically for variable component amplitudes, implying also on variable latencies. The analytical qualities are compared to simulated results confirming the estimator's capacity to track morphological changes of simulated signals. Depending on the extent of allowed variations, the estimator enables extracting EP's with signal to noise ratios as low as 15 dB. The performance of the estimator is demonstrated on single movement related brain potentials, recorded during changing experimental conditions and thus expected to yield varying EP's. The results presented show that single MVEP's can be successfully estimated from the noisy background activity, which enables tracking of transient qualities of brain responses throughout an experimental session. View full abstract»

Functional correlates with leftright asymmetry of visual evoked potentials in stroke patientsmodeling and experimental results
Page(s): 1.2.2/1  1.2.2/3The correlation between the FIM (Functional Independence Measure) on admission and the leftright scalp potential amplitude was studied in poststroke patients, using flash Visual Evoked Potential (VEP) stimulation and theoretical model of the head. It was found that leftright asymmetry of the scalp VEP may be useful indices for rehabilitation in patients after brain damage. The modeling study shows that leftright asymmetry of the scalp VEP in patients after brain damage may be a result of changes in the conductivity of the damage region. View full abstract»

A comparison of neural network and Bayes recognition approaches to the diagnostic of multiple sclerosis
Page(s): 1.2.3/1  1.2.3/5This article describes the application of a MultiLayer Perceptron (MLP) to the problem of diagnosing Multiple Sclerosis (MS). The classification information is obtained by a Trigeminal Evoked Potential (TEP) test. The performance of the MLP is compared with that of the human experts and the Bayes classifier. The efficiency of the neural network and the classical classifiers in conjunction with 4 types of features  the Fourier transform (FT), the peak position, the ARX model coefficient and the temporal wave form  are examined. Although a large clinical data base would be necessary, before this approach can be fully validated, the initial results are very promising. The MLP was found to be less susceptible to the number of features used. The ability of the MLP classifier to generalize is far better than that of the Bayes classifier. View full abstract»

Clustering of nonconvex patterns for medical imaging
Page(s): 1.2.4/1  1.2.4/5The study treats the problem of partitional clustering of nonconvex patterns of arbitrary shape, with no definite analytical description of the data. A new method for calculating the distance is defined, according to a data induced metric principle. Calculation of the distance is carried out using weighted graphs. The new distance is introduced in the fuzzy kmeans algorithm, and the modified fuzzy kmeans algorithm is tested on synthetic data, and on MRI images of the heart. View full abstract»

Diskbased videoondemand storage servers: requirements, challenges and (some) solutions
Page(s): 1.3.1/1  1.3.1/5A videoondemand (VOD) storage server is a parallel, storagecentric system used for playing a large number of relatively slow streams of compressed digitized video and audio concurrently. Data is read from disks in relatively large chunks, and is then "streamed" out onto a distribution network. The primary design goal is to maximize the ratio of the number of concurrent streams to system cost while guaranteeing glitchfree operation. This paper characterizes the VOD applications and then identifies several important issues along with an outline of possible approaches to dealing with them. Issues include the accommodation of unlimited demand for concurrent "private" viewing of the few hottest movies with limited resources, multizone recording and the resulting variable disk transfer rate, as well the interplay between faulttolerance, load balancing, the size of RAM buffers and the organization of the storage subsystem. View full abstract»

A framework for type system definition and manipulation in C++
Page(s): 1.3.2/1  1.3.2/5Metaprogramming sets its goal in treating programs as data objects, and as such, manipulating and modifying them. We describe a framework that provides an essential part of a metaprogramming environment, namely types. We have implemented an extendible class library in C++ that presents types as first class values, and uses a simple syntax to allow easy definition, creation and manipulation of types. This poses a framework which can be used in several application domains, including language processing tools, database applications, and CASE tools. View full abstract»

Visualizing programs using graphs
Page(s): 1.3.3/1  1.3.3/4Graphs seem to be a natural way to represent various aspects of program execution. And therefore, graphs are commonly used by software visualization to display, interpret, document and convey the static and dynamic nature of program execution. The major obstacles which must be considered before graphs can be used in software visualization tools, are the conceptual problems, that is, the difficulty of the interpretation of a graph by a user, and the display problems, that is, the difficulties in handling large graphs and drawing dynamically changing graphs on the display. In this paper, we show a set of techniques which together overcome the complexity of large graphs, their display, and the handling of the dynamic changes in graphs. We demonstrate these techniques based on a variety of tools which were developed in the past. View full abstract»

A verification scheme for pipelined vector architectures
Page(s): 1.3.4/1  1.3.4/5As a means for improving performance, advanced vector processors use an extension of pipelining, called vector chaining, whereby the execution of independent instructions is overlapped. The complexity of vector chaining architectures, together with their inherent parallelism and asynchrony, renders their verification extremely difficult. This paper presents an efficient simulationbased scheme for verifying such architectures. The scheme presented here can be applied to other verification problems where the simulation expected results are not deterministic but belong to set a of computable possibilities. View full abstract»

The Parallel Booster: software architecture and optimization results
Page(s): 1.3.5/1  1.3.5/4The Parallel Booster is a software package to enable the development of parallel programs literally in a couple of months, and to allow full runtime exploitation of the installed computer base by means of a set of unique functions. The architecture of the Booster and some of these functions are succinctly described. As a concrete proof of the Booster power, the parallelization of an actual application is discussed. Both claims were demonstrated: dramatic acceleration of an optimization computation was achieved. The modifications were of a surgical nature, explaining why short development efforts are possible. View full abstract»

A study of finite size power isotropic radiators
Page(s): 1.4.1/1  1.4.1/5A class of localized current sources that radiate power isotropically is presented. Their farfields are given as well as the symmetries of the associated polarization distributions, and, in addition, the multipole expansion of the fields of these radiators is calculated. View full abstract»

A numerical absorbing boundary condition for finite element analysis
Page(s): 1.4.2/1  1.4.2/5In this paper we present a novel approach to deriving local boundary conditions that can be employed in conjunction with the Finite Element Method (FEM) to solve electromagnetic scattering and radiation problems. The proposed Numerical Absorbing Boundary Condition (NABC) approach can he applied to both the nodal and the edgebased formulations of the FEM. The key step in the NABC approach is to derive linear relationships that link the values of the field at the boundary grid points or the field along the boundary edges, depending on formulation, to those at the neighboring ones. Derived via a robust SVDbased scheme, these linear relationships are satisfied, to within a certain tolerance, by all of the outgoing wave components. They can thus be used in lieu of the FEM equations employing the usual absorbing boundary conditions for the boundary edges. The present approach not only provides a simple way to derive an accurate boundary condition for mesh truncation at an arbitrary frequency, but also preserves the banded structure of the FEM matrices. Furthermore, the validity of the method is demonstrated at very low frequencies where the conventional absorbing boundary conditions are prone to fail. View full abstract»

Improving impedance matrix localization by a digital filtering approach
Page(s): 1.4.3/1  1.4.3/5Wavelet bases have been employed recently in numerical solutions of integral equations encountered in electromagnetic scattering problems. The advantage of these bases lies in their ability to render the problem impedance matrix more localized. In this paper, we propose a digital filtering approach for integrating wavelet transforms into existing electromagneticscattering numericalsolvers. The suggested approach facilitates a much faster implementation of the transform. It also allows the incorporation of other ideas such as wavepackets and bestbasis from the discipline of digital signal processing. A physical interpretation of the basis functions obtained by a few selected structures of digital filters is given, and their usefulness is explained. A numerical example is given for the case of TM scattering by a square cylinder. View full abstract»

Circular polarization microstrip array on a conical surface
Page(s): 1.4.5/1  1.4.5/3Antenna elements based on microstrip technology are low profile, conformal, low weight and low cost, as such they are commonly used in many applications. Lately, extensive effort was devoted to compute the radiation characteristics of microstrip elements on flat and cylindrical surfaces, but only minor attention has been given to the analysis of microstrip elements on a conical surface. Such a geometry has a significant importance in satellite and space communication. In this study we utilized the cavity model (1979) to compute the input impedance of the patch. We used spherical wave functions expansion to compute the internal cavity fields and the external radiated fields. The circular polarization of the microstrip element was obtained by exciting two orthogonal modes with 90 deg. phase delay among them. To obtain an omnidirectional azimuth, coverage, three microstrip elements were interconnected and equally spaced along the cone surface. The antenna was built and tested and good agreement between the computed and measured results was obtained. View full abstract»

Matrix thinning with reduced field testing (RFT)
Page(s): 1.4.6/1  1.4.6/4A number of methods have been proposed recently for the generation of sparse, rather than dense matrices in the context of integral equations. Most notably, we refer to Canning's impedance matrix localization (IML), the wavelet expansion, the fast multipole method (FMM), and the use of unidirectional basis functions, including a dual source representation and the complex multipole beam approach. A good approximation for a sparse matrix is obtained by the reduced field testing (RFT) method, presented here. The method is a spatial domain alternative to the spectral domain FMM, based on the distinction between near field and far field interactions, and providing a single, explicit sparse matrix invertible by either direct or iterative methods. The procedure does not change essentially the way by which MoM matrix elements are computed, hence it can be incorporated into existing MoM codes. Accuracy has been demonstrated by a number of examples. View full abstract»

Digital simulation of transient phenomena in power transformers
Page(s): 1.5.1/1  1.5.1/5The switching of a power transformer to a source (electric power system) from one side leads to the occurrence of inrush currents. Typical for these currents when all other windings of the transformer are opencircuited is the following: the peak values are higher than the rated currents, the curves are highly unsymmetrical and diminish faster than the shortcircuit currents. These characteristics are used by the transformer relay protections to distinguish between the shortcircuit currents and the inrush currents. View full abstract»

Domains of ferroresonance occurrence in voltage transformers with or without damping reactors
Page(s): 1.5.2/1  1.5.2/5This paper is aimed at investigating the circumstances under which ferroresonance is liable to occur in a 400 kV Gas Insulated Substation (GIS) equipped with inductive voltage transformers, and at establishing the efficiency of damping reactors connected to the secondary winding of these transformers in order to improve the damping of the ferroresonance or to avoid its appearance. The paper outlines the computational model. Various operational schemes liable to be driven into ferroresonance are taking into account. The ferroresonance occurrence domains, determined using the computational model, emphasize that the damping reactor efficiency depends on the value of the external serial resistor and on the various operational schemes. View full abstract»

Analysis of a photovoltaic water pumping system
Page(s): 1.5.3/1  1.5.3/5A solar powered photovoltaic water pumping system has been installed and field tested at James Cook University of North Queensland, Australia. The results of the experiment are discussed based on field measurements and computer modeling of the various system components in order to investigate the efficiency of each component and the performance of the entire system. View full abstract»

Superconducting inductive current limiter as a power system component
Page(s): 1.5.4/1  1.5.4/5One promising option for the current limiter is a device based on a fast transition of a superconducting material from the superconducting state with zero resistance to the normal state with pronounced resistance. The superconducting current limiting device being developed by us is based on an idea of inductive coupling of an active high critical temperature superconducting element to the power network. We present the results of theoretical investigation of the device's influence on the electromagnetic processes in the circuit to be protected. We have performed a simulation of a 13.8 kV device performance with the use of a mathematical model that takes into account both electromagnetic and thermal processes in the device. We have introduced the inhomogenities of the ring properties into the model and investigated their influence on the transient characteristics of the device. The basic properties of the current limiter were experimentally investigated using 50 V/10 A and 1000 V/25 A prototypes. View full abstract»