Spl Rho Research Articles

Accuracy analysis for wavelet approximations

"Constructive wavelet networks" are investigated as a universal tool for function approximation. The parameters of such networks are obtained via some "direct" Monte Carlo procedures. Approximation bounds are given. Typically, it is shown that such networks with one layer of "wavelons" achieve an L(2) error of order O(N(-(rho/d))), where N is the number of nodes, d is the problem dimension and rho is the number of summable derivatives of the approximated function. An algorithm is also proposed to estimate this approximation based on noisy input-output data observed from the function under consideration. Unlike neural network training, this estimation procedure does not rely on stochastic gradient type techniques such as the celebrated "backpropagation" and it completely avoids the problem of poor convergence or undesirable local minima.

Transform-domain adaptive filters: an analytical approach

Transform-domain adaptive filters refer to LMS filters whose inputs are preprocessed with a unitary data-independent transformation followed by a power normalization stage. The transformation is typically chosen to be the discrete Fourier transform (DFT), although other transformations, such as the cosine transform (DCT), the Hartley transform (DHT), or the Walsh-Hadamard transform, have also been proposed in the literature. The resulting algorithms are generally called DFT-LMS, DCT-LMS, etc. This preprocessing improves the eigenvalue distribution of the input autocorrelation matrix of the LMS filter and, as a consequence, ameliorates its convergence speed. In this paper, we start with a brief intuitive explanation of transform-domain algorithms. We then analyze the effects of the preprocessing performed in DFT-LMS and DCT-LMS for first-order Markov inputs. In particular, we show that for Markov-1 inputs of correlation parameter /spl rho//spl isin/[0,1], the eigenvalue spread after DFT and power normalization tends to (1+/spl rho/)l(1-/spl rho/) as the size of the filter gets large, whereas after DCT and power normalization, it reduces to (1+/spl rho/). For comparison, the eigenvalue spread before transformation is asymptotically equal to (1+/spl rho/)/sup 2//(1-/spl rho/)/sup 2/. The analytical method used in the paper provides additional insight into how the algorithms work and is expected to extend to other input signal classes and other transformations. >

Magnetic and electrical properties of NiFe-oxide mixture films

In order to improve the performance of magnetoresistive heads, higher saturation flux density Bs and higher resistivity /spl rho/ are preferable for Soft Adjacent Layer (SAL). NiFe-oxide (oxide=SiO/sub 2/, Al/sub 2/O/sub 3/, ZrO/sub 2/ and Ta/sub 2/O/sub 5/) mixture films were investigated for SAL. Among these films, the NiFe-ZrO/sub 2/ system is preferable because of its high Bs /spl rho/ and low coercivity. The stability of the NiFe-ZrO/sub 2/ system at 250/spl deg/C was also investigated, which shows that Bs and /spl rho/ change by annealing and the changes increase with ZrO/sub 2/ content. MR devices of 3.2 /spl mu/m width and 2.8 /spl mu/m height were fabricated and transfer curves were measured. The maximum output voltage of the device in which NiFe-12mol% ZrO/sub 2/ was employed was 23% larger than the calculated value for the device in which NiFe-Nb (Bs /spl rho/=60 Tesla /spl mu//spl Omega/cm) was assumed to be employed.

Soft magnetic properties and microstructure of nanocrystalline Fe-Hf-N sputtered films

Magnetic properties, structure and electrical resistivity (/spl rho/) of Fe-Hf-N films with a compositional range of 10-17 at.% Hf and 9-28 at.% N were investigated. In an as-deposited state, the high permeability (/spl mu/') above 700 at 10 MHz is obtained in the compositional range of about 12-14 at.% Hf and about 9-19 at.% N. The highest /spl mu/' of 1200 is achieved for Fe/sub 72/Hf/sub 13/N/sub 15/ film consisting of the nanoscale BCC grains and the amorphous phase with the high Curie temperature (T/sub c/) due to higher contents of Hf and N than those in the BCC phase. The soft magnetic properties are presumably caused by the reduced effective anisotropy resulting from the effect of the nanoscale BCC grains and the rather high magnetization of the amorphous phase at room temperature which does not inhibit the exchange coupling between the grains. Fe/sub 65/Hf/sub 11/N/sub 24/ film with the mixed structure exhibits the saturation magnetic flux density (B/sub 3/) of 1.2 T and the flat /spl mu/' characteristics of 1000 up to 100 MHz after annealing at 673 K for 3 hrs under an uniaxial field. This excellent Cc characteristics originates in the rather large magnetic anisotropy field (H/sub k/) of 510 A/m and the high /spl rho/ of 2.7 /spl mu//spl Omega/m m which comes from the Hf and N-riched amorphous phase.

Routing and scheduling in heterogeneous systems: a sample path approach

Consider the problem of routing customers to a set of K parallel servers that have different rates. Each server has a buffer with infinite capacity. The arrival process is general and the service times are assumed to be i.i.d. exponential random variables. Using sample path arguments, we show that, given any Bernoulli policy /spl pi/, there exists another policy /spl rho/ which outperforms /spl pi/ by partially using a randomized version of a round-robin policy. Moreover, /spl rho/ is easily specified and implemented. We present extensions of our results to systems with finite capacities and service times that have an increasing hazard rate. Finally, a similar result is shown to hold in the context of scheduling customers from a set of K parallel queues.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Generating and approximating nondominated coteries

A coterie, which is used to realize mutual exclusion in a distributed system is a family C of incomparable subsets such that every pair of subsets in C has at least one element in common. Associate with a family of subsets C a positive (i.e., monotone) Boolean function f/sub c/ such that f/sub c/(x)=1 if the Boolean vector x is equal to or greater than the characteristic vector of some subset in C, and 0 otherwise. It is known that C is a coterie if and only if f/sub c/ is dual-minor, and is a nondominated (ND) coterie if and only if f/sub c/ is self-dual. We introduce an operator /spl rho/, which transforms a positive self-dual function into another positive self-dual function, and the concept of almost-self-duality, which is a close approximation to self-duality and can be checked in polynomial time (the complexity of checking positive self-duality is currently unknown). After proving several interesting properties of them, we propose a simple algorithm to check whether a given positive function is self-dual or not. Although this is not a polynomial algorithm, it is practically efficient in most cases. Finally, we present an incrementally polynomial algorithm that generates all positive self-dual functions (ND coteries) by repeatedly applying p operations. Based on this algorithm, all ND coteries of up to seven variables are computed.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Polynomial-time construction of codes .II. spherical codes and the kissing number of spheres

A spherical code is a finite set X of points lying on the unit sphere of R/sup n/. For such a set, we define /spl rho/(X) as the minimum of the squared distances /spl par/x-y/spl par//sup 2/, when x, y/spl isin/X and x/spl ne/y. Define R(/spl rho/)=lim sup n/spl rarr//spl infin/, /spl rho/(X)=p log/sub 2/CardX/n. Chabauty in 1953 and Shannon in 1959 have given a lower bound for R(/spl rho/), namely, R(/spl rho/)>R/sub CS/(/spl rho/)=1-1/3log/sub 2//spl rho/(4-p). The complexity of construction of the spherical codes used in order to get this bound is doubly exponential. The polynomially constructible spherical bound R/sub pol/(/spl rho/) is defined as above with the additional restriction that only families of codes with polynomial complexity of construction are considered. We prove R/sub pol/(/spl rho/)/spl ges/R/sub CS/(/spl rho/)/2, if /spl rho//spl les/1.535. Denote by /spl tau//sub X/(n) the number of spheres of equal radius that touch one sphere in the n-dimensional space given by some explicit family X, that is, a family of arrangements of spheres). The asymptotic polynomially constructible kissing number is /spl theta//sub pol/=lim sup(log/sub 2//spl tau//sub X/(n))/n, when X ranges over all polynomially constructible families. We prove /spl theta//sub pol//spl ges/2/15=0.133/spl middot//spl middot//spl middot/. >

Use of copolar correlation coefficient for probing precipitation at nearly vertical incidence

Presents observations of the copolar correlation coefficient between horizontally and vertically polarized echoes |/spl rho//sub h/spl nu//(0)|. These were made with ground-based and airborne weather radars at nearly vertical incidence. A sharp decrease of |/spl rho//sub h/spl nu//(0)| occurs at the bright band bottom, and is attributed to a varying mixture of hydrometeors with diverse shape, size, and thermodynamic phase. The largest contribution to decorrelation seems to come from wet aggregates; this is substantiated by consideration of two simple models. One consists of randomly oriented wet prolate spheroids, and the other considers an ensemble of distorted spheres. Prolates with axis ratios of 3 or distorted spheres with rms roughness equal to 15% of the diameter decrease the correlation to 0.8 at S band. At Ku band and for the size range encountered in the bright band, the decrease is a function of equivalent diameter because scattering is in the Mie regime. |/spl rho//sub h/spl nu//(0)| measurement at 13.8 GHz and from the aircraft are the first ever. Also, differential phase and differential reflectivity at a 10/spl deg/ off nadir are the first of its kind. These last two variables showed a distinct signature in the bright band. This is significant because it might lead to applications on airborne or spaceborne platforms.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Detection of forests using mid-IR reflectance: an application for aerosol studies

The detection of dark, dense vegetation is an important step in the remote sensing of aerosol loading. Current methods that employ the red (0.64 /spl mu/m) and the near-IR (0.84 /spl mu/m) regions are unsatisfactory in that the presence of aerosols in the scene distorts the apparent reflectance in the visible and near-IR ranges of the spectrum. The mid-IR spectral region is also sensitive to vegetation due to the absorption of liquid water in the foliage, but is not sensitive to the presence of most aerosols (except for dust). Therefore, mid-IR channels on the AVHRR and EOS-MODIS (e.g., the 3.75 /spl mu/m or the 3.95 /spl mu/m channels) have a unique potential for the remote sensing of dark, dense vegetation, particularly in the presence of biomass burning smoke or industrial/urban haze. The reflective part of the 3.75 /spl mu/m channel (/spl rho//sub 3.75/) is applied to images of the AVHRR over the eastern United States. This channel was found to be correlated to reflectance at 0.64 /spl mu/m (/spl rho//sub 0.64/), less sensitive to haze than the visible channel and superior to both the 0.64 /spl mu/m reflectance and the normalized difference vegetation index (NDVI) to determine forest pixels in an image. However, its application to monitor the seasonal evolution of vegetation is presently questionable. For the purpose of the remote sensing of aerosol over dark, dense vegetation, it is proposed that the dark, dense vegetation be determined from /spl rho//sub 3.75/<0.025. These findings may have further implications for other specific applications of the remote sensing of vegetation in hazy atmospheres.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Proposed expansions for the capacitance of a square centered in a circle

Expansions for the capacitance per unit length of a square centered in a circle are proposed: one in terms of, /spl rho/, the ratio of the diameter of the square to the diameter of the circle and the other in terms of /spl delta/=1-/spl rho/. Limits on the accuracy of the expansion in terms of /spl rho/ are given in graphical form.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

A self-consistent characterization methodology for Schottky-barrier diodes and ohmic contacts

Based on the simple interfacial-layer theory, the extraction methods for the interface parameters of the metal-semiconductor contact have been developed and applied to characterize both the Schottky-barrier diodes and the ohmic contacts in a self-consistent manner. It has been shown that the physical parameters at the metal-semiconductor interface can be extracted from the I-V characteristics of the Schottky-barrier diodes and the degradation of the thermal-equilibrium barrier height due to the thermal cycle can be directly modeled in terms of the extracted interface parameters. Besides, using the extracted parameters, the specified surface-treatment process can be evaluated by the extracted thermal-equilibrium barrier height, and thus the strongly process-dependent specific contact resistivity /spl rho//sub c/ of the ohmic contacts can be theoretically calculated by a modified tunneling model considering the impurity band. Furthermore, by comparing the simulated results and the measured /spl rho//sub c/ data deduced from the Al and Ti contacts on both doping types of the Si-substrate, satisfactory agreements have been obtained.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Optimum loop filter in hybrid coders

Examines the role of the loop/interpolation filter in the motion compensation loop of hybrid coders. Using the Wiener filtering concept and the statistical model developed by Chen and Pang (1992), the optimum filter that minimises the prediction error has been found. The result is expressed in an explicit form in terms of a correlation parameter, /spl rho/ and an inaccuracy parameter, /spl alpha/. It explains many current practices in MPEG and H.261 coders, as well as the leakage predictor, 3-tap versus 8-tap filters and other related issues. The analysis shows that minimum bit rate can only be achieved if the loop filter matches the statistical characteristic of the motion-compensated signal. Furthermore, since the motion noise characteristic could be very different in the horizontal and vertical direction for many sequences, the decision to deploy the optimum filter should be made separately in the two directions. The paper also derives the scalar loop filter in DCT domain. The scalar filter is sub-optimal, but it requires less computational load than the spatial domain filter (64 versus 484 multiplications per 8/spl times/8 block). Experiments show that it performs almost as efficiently as the optimum 3-tap spatial domain filter, thus ascertaining that its performance has not been significantly compromised by the scalar requirement. Experimental simulations on test sequences confirm the theoretical optimum results, and indirectly show that the simple statistical model used in the derivation is adequate.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Corrective memory by a symmetric sparsely encoded network

A neural network that retrieves stored binary vectors, when probed by possibly corrupted versions of them, is presented. It employs sparse ternary internal coding and autocorrelation (Hebbian) storage. It is symmetrically structured and, consequently, can be folded into a feedback configuration. Bounds on the network parameters are derived from probabilistic considerations. It is shown that when the input dimension is n, the proportional activation radius is /spl rho/ and the network size is 2/sup /spl nu/n/ with /spl nu/>1-h/sub 2/(/spl rho/), the equilibrium capacity is at least 2/sup /spl alpha/n//8n/spl rho/(1-/spl rho/) for any /spl alpha/<1-h/sub 2/(/spl rho/), where h/sub 2/(/spl middot/) is the binary entropy. A similar capacity bound is derived for the correction of errors of proportional size /spl rho/ or less, when /spl rho//spl les/0.3. The performance of a finite-size symmetric network is examined by simulation and found to exceed, at the cost of higher connectivity, that of the Kanerva (1988) model, operating as a content addressable memory.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Giant magnetoresistance in actinide systems

In contrast to ferromagnetic uranium intermetallics, which exhibit a large drop below the ordering temperature, the resistivities of uranium antiferromagnets stay at much larger values below T/sub N/. In this contribution, we demonstrate that a large portion of the residual resistance of uranium antiferromagnets can be removed by driving the compounds through metamagnetic transition to a field-aligned state. The consequence is the field-induced drop of resistance, which can exceed /spl Delta//spl rho///spl rho/(0T)=90%.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Giant magnetoresistance and antiferromagnetically coupled Co fraction in Co/Cu multilayers with varying number of periods

The giant magnetoresistive signal of Co/Cu multilayers as a function of the number N/sub Co/ of Co layers is experimentally investigated for Cu thicknesses of 1 nm, i.e. in the 1st antiferromagnetic maximum, and for Co layer thickness of 1 nm. The stacks were deposited by DC magnetron sputtering on an Fe buffer. The fraction F/sub AF/ of the film that couples antiferromagnetically is derived from the hysteresis curves. A correction of F/sub AF/ is required for the buffer and one Co layer. The F/sub AF/ of stack periods close to the buffer is about 1, while F/sub AF/ reduces at large stack thicknesses. The GMR of stacks with small thicknesses is limited by the diffuse scattering at the outer surfaces. The GMR of thicker stacks depends on F/sub AF/ and the resistivity. Furthermore, the meaning of F/sub AF/ is discussed. By optimizing N/sub Co/, we achieved /spl Delta//spl rho///spl rho/'s as high as 72% at room temperature. >

Weighted backprojection approach to cone beam 3D projection reconstruction for truncated spherical detection geometry

A new analytical three-dimensional cone beam reconstruction algorithm is presented for truncated spherical detection geometry. The basic idea of the proposed algorithm is the formation of spatially invariant 3D blurred back-projected volumetric image by the use of the weighted backprojection of cone beam projection data and subsequent 3D filtering using an acceptance angle dependent rho filter. The backprojection weighting function is calculated on the basis of each given geometrical condition, i.e. detection geometry or degree of truncation, position of cone beam apex, and backprojection point. The proposed algorithm is derived analytically and is computationally efficient. Performance of the algorithm is evaluated by the reconstruction of 3D volumetric images using simulated data from arbitrarily truncated spherical detector geometries.

Received signal-level characteristics in a wide-band mobile radio channel

A mobile propagation model aimed at clarifying fundamental propagation characteristics in received signal-level variation for wide-band transmission is proposed. On the basis of this model, the author derives an expression for a received signal level in wide-band transmission and examines the fundamental signal-level characteristics by computer simulation and experiment. Both simulation and measurement results agree well and the results follow. For a received signal-level variation in wide-band transmission: first, received signal-level depth becomes shallower with increasing receiver bandwidth 2/spl Delta/f, and the level has no Rayleigh distribution. In an urban area when 2/spl Delta/f is 3 MHz, the level difference between the cumulative probability 50% and 1% values is about 5 dB. Second, received signal-level distribution depends on the number of arriving waves N and path length difference |/spl Delta/L/sub ij/|. The level depth becomes shallower with increasing N and |/spl Delta/L/sub ij/|. Third, received signal-level distribution is almost independent of radio frequency f/sub c/. The author also derives expressions for the autocorrelation coefficient /spl rho/(z) and frequency coefficient /spl rho/(s) of the received signal level. /spl rho/(z) is independent of 2/spl Delta/f, and is about J/sub 0/(Z//spl lambda//sub c/)/sup 2/, which is known as narrow-band reception. The /spl rho/(s) becomes higher with increasing 2/spl Delta/f. >

Size and self-field effects in giant magnetoresistive thin-film devices

Giant magnetoresistance (GMR) was measured as a function of device size for patterned NiCoFe/Cu and NiFe/Ag films. For the quasi-granular NiCoFe/Cu films, the normalized maximum change in resistivity /spl Delta//spl rho///spl rho/ was 8% for most of the samples. For the NiFe/Ag films, antiparallel alignment was achieved through magnetostatic coupling, not exchange fields, with a /spl Delta//spl rho///spl rho/ of 4.5%. The films were patterned into stripes with Au current leads for size-effect measurements. The height of the stripes varied from 0.5 to 16 /spl mu/m and the track width varied from 1 to 16 /spl mu/m. Discrete switching events and anomalous low-field dips in the response were observed for both materials for small device sizes. Self-field and heating effects due to the applied current were investigated for the NiFeCo/Cu films. The effect of the self-field produced by the applied current was separated from the thermal contribution and was found to reduce the response by over 32% for a current density of 10/sup 7/ A/cm/sup 2/.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Improved eddy-current decay method for resistivity characterization

Eddy-current decay is a unique, nondestructive method for determining the low-temperature resistivity of large samples of pure metal. Furthermore, it is the only means available for measurement of the residual resistivity ratio (RRR), /spl rho/(273 K)//spl rho/(4 K) of samples with shapes that do not lend themselves to conventional four-wire resistance measurement techniques. An improvement to an earlier implementation of the eddy-current decay method of resistivity characterization is presented. It involves modernizing the earlier apparatus by the use of a digitizing oscilloscope, commercial curve-fitting software, digital averaging techniques, and modern electronics. Data are shown for high-purity copper cylinders.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Inverter performance of 0.10 μm CMOS operating at room temperature

The switching performance of 0.10 /spl mu/m CMOS devices operating at room temperature has been discussed on the basis of both experimental and simulated results. The measured propagation delay time of a 0.10 /spl mu/m gate length CMOS has been quantitatively divided into intrinsic and parasitic components for the first time. The results have shown that the drain junction capacitance strongly affects the propagation delay time in the present 0.10 /spl mu/m CMOS. The switching performance of a 0.10 /spl mu/m ground rule CMOS has been simulated by using device parameters extracted from the experimental results. In the 0.10 /spl mu/m ground rule CMOS, it has been shown that an increase of the contact resistance will degrade the propagation delay time, which is one of the most essential problems in further device miniaturization. It has been also demonstrated that even if the specific contact resistance /spl rho//sub c/ is reduced to be less than 1/spl times/10/sup -7/ /spl Omega/ cm, further reduction of the gate overlap capacitance C/sub ov/ will be required to achieve the propagation delay time to be less than 10 ps in the 0.10 /spl mu/m ground rule CMOS at room temperature. >