Search

All circuit components have uncertainties inherent to the underlying fabrication process that, in large networks, may change the overall circuit response unpredictably. In the absence of a robust general model to incorporate individual component uncertainties and the concomitant stochastic thermal noise characteristics, under-simulating from the massively high-dimensional experiment space via Monte Carlo techniques and over-designing the final product against potentially undiscovered faults have become de facto standard. Such practices do not only cost simulation time, circuit area, and power, but also provide a partial understanding of the underlying uncertainty and of ways to exploit it. This paper investigates the impact of component uncertainties in linear resistive networks, where individual elements are subject to Johnson-Nyquist noise.

Distributed networks comprising decision-making units of variable competence are used to perform a sequence of tasks. Estimation of the local competences based solely on comparisons of local decisions on dynamically changing tasks is necessary for improving decision aggregation performance. This paper addresses distributed estimation of local competences on networks of heterogeneous decision-making units. We propose an order-preserving metric that depends on the local network structure to measure the competence of each node in the network instantaneously. We investigate two regimes of operation: First, the tasks are received at low frequency, or sequentially, hence the network can reach a consensus in-between tasks. Second, tasks are received at high frequency, or in batches, in which case, the network reaches to a consensus after tasks are processes locally.

Existing wheelchair control interfaces, such as sip & puff or screen based gaze-controlled cursors, are challenging for the severely disabled to navigate safely and independently as users continuously need to interact with an interface during navigation. This puts a significant cognitive load on users and prevents them from interacting with the environment in other forms during navigation. We have combined eyetracking/gaze-contingent intention decoding with computer vision context-aware algorithms and autonomous navigation drawn from self-driving vehicles to allow paralysed users to drive by eye, simply by decoding natural gaze about where the user wants to go: A.Eye Drive. Our "Zero UI" driving platform allows users to look and interact visually with at an object or destination of interest in their visual scene, and the wheelchair autonomously takes the user to the intended destination, while continuously updating the computed path for static and dynamic obstacles. This intention decoding technology empowers the end-user by promising more independence through their own agency.

Computational units implemented on nanoscale physical substrates are susceptible to errors that can be catastrophic if not mitigated. Statistical error compensation techniques have become prevalent to safeguard computational units against such hardware-failures. Algorithmic Noise Tolerance (ANT) is one such technique that utilizes a low-fidelity replica unit to detect and bypass such failures occurring within the primary (main) computational unit. Connections between ANT and the binary hypothesis testing as well as the information theoretic CEO problem have been explored for sub-exponential error profiles, quadratic and logarithmic distortion functions. However, there exist fundamental performance limits of ANT approach even without such model-dependent restrictions. The purpose of this paper is to explore fidelity-dependent conditions that are universal over the statistical properties of the computational units under which, the overall performance of ANT is arbitrarily close to the fundamental limits.

In this paper, our goal is to compare performances of three different algorithms to predict the ratings that will be given to movies by potential users where we are given a user-movie rating matrix based on the past observations. To this end, we evaluate User-Based Collaborative Filtering, Iterative Matrix Factorization and Yehuda Koren's Integrated model using neighborhood and factorization where we use root mean square error (RMSE) as the performance evaluation metric. In short, we do not observe significant differences between performances, especially when the complexity increase is considered. We can conclude that Iterative Matrix Factorization performs fairly well despite its simplicity., Comment: This work has been accepted to the 2017 IEEE ICMLA

Sufficient conditions for periodic reference signals to ensure lossless natural sampling of finite energy, band-limited signals are derived in the context of pulse-width modulation (PWM). For sawtooth reference signals, which are commonly used in PWM generation, lossless natural sampling conditions in the Landau sense are characterized in terms of the range and period of these signals. The results are extended to any absolutely continuous reference signal that spans its dynamic range monotonically on a pulse interval while satisfying mild derivative conditions. Finally, lossless natural sampling conditions for double-sided reference signals are proposed, revealing a design initiative to reduce PWM generator complexity.

Pulse width modulation is discussed in a comparator with periodic reference signal setting with the emphasis on distinctions between natural and uniform sampling. Motivated by low generator complexity of natural sampling, the convergence characteristics between natural and uniform sampling of a continuous, finite energy, band-limited signal is investigated. The necessary and sufficient conditions for lossless natural sampling are derived and shown to ensure lossless uniform sampling as well. A geometrical framework is constructed to upper bound the deviation between uniform and natural samples and the corresponding sampling instances. Finally, the relation between deviation energy and oversampling factor is discussed.

This paper introduces the proof of concept of low-Earth-orbit satellite attitude determination using GNSS, mainly GPS satellite signals. Two methods are considered and compared in detail: carrier phase measurements and signal-to-noise ratio (SNR-RSS) method. A preliminary design to be used in testing the algorithms is explained, and results of simulations with RSS method are presented.

A performance bound on low-pass reconstruction of a finite energy, band-limited signal from its corresponding pulse width modulated (PWM) signal is given. A finite dimensional input signal structure is proposed and its characteristics such as ℒ ∞ norm and convergence rate in the tail regions are investigated. The criteria on sampling rate and reference signal structure for lossless PWM generation are specified. With uniform oversampling, different PWM structures are constructed using a generator consisting of an ideal sampler and a comparator. An equivalent model for PWM generation and low-pass reconstruction is proposed in order to separate infinite energy components in the modulated signal from the finite energy information bearing components. Using the equivalent model, bounds on distortion energies due to low-pass reconstruction and the corresponding signal to distortion ratios are given. Characteristics of low-pass reconstruction from different PWM signals such as output shifts, blurring effects and distortion attenuations are discussed.

A stochastic analysis for pulse width modulation (PWM) is given. Under a discrete random generator process with independent identically distributed (i.i.d.), continuous, range-limited samples, stochastic models for different PWM signals are proposed. Using randomization of the signal starting point, wide sense stationarity (WSS) of proposed PWM signals are shown. Autocorrelation functions and their corresponding power spectrum densities (PSD) are derived in terms of the modified complementary cumulative distribution function of i.i.d samples. For the case where the samples are i.i.d. uniformly distributed, the proposed autocorrelation functions are tested with simulations.

Modernization of enterprise systems creates new challenges to corporations on how to transform the existing IT assets into service-oriented architectures. Exposing existing system functionality in the form of enterprise services is one efficient way to leverage the value and reusability of enterprise systems. In this respect, existing system components can be used as the foundation of newly created services. However, one problem to overcome is the lack of system semantics to support the reuse of existing source code. In this paper, we present an approach to discover semantic relations between high-level business functionalities and low-level source code components. Our approach is based on the idea that the gap between service components and source code components of an enterprise system can be bridged by mapping through entity similarity. The procedure of our approach consists of two steps; (1) generating ontologies of source code components and the proposed enterprise services, (2) semantic mapping of ontology entities through similarity measures. We evaluate our approach in the framework of a commercial enterprise systems application. Initial results indicate that the proposed approach is useful for discovering service- related knowledge from existing source code components.

Automotive short-range sensor networks at 24 GHz represent the first truly high volume commercial system at mm-wave, with first year production quantities already eclipsing previous products such as autonomous cruise control (ACC), and the various microwave radio networks (e.g. LMDS). The emphasis on product quality, reliability, cost, and performance demanded by automotive OEMs has meant that an innovative approach to product development, packaging, and test has had to be introduced to mm-wave products. This paper will highlight some of the key technical hurdles and their solutions in the development and introduction to production of a 24 GHz UWB radar sensor based on custom-designed SiGe ICs.

A multimode TRL (MTRL) calibration technique is discussed to characterize broadband differential devices. This algorithm has the advantage of using ground-signal-signal-ground (G-S-S-G) probes in the measurements without the frequency limitations of some other 4-port algorithms due to the coupling between the signal lines. Calibration standards and the issues associated with them are described. Practical issues of the underlying theory, such as assigning the calculated eigenvalues to the correct mode and to the propagation direction, have also been addressed and one case-specific solution is suggested. Also an MTRL algorithm has been adopted to avoid some of the difficulties of calibration structures with MTRL. Odd mode test results of differential active circuits with MTRL using G-S-S-G probes are shown to be consistent with the two port test results of the same devices using S-G probes.

A fully integrated, plastic packaged, 24 GHz SiGe receiver chip is presented. The chip has been manufactured using a commercially available SiGe foundry process. It can be used in a variety of applications including automotive radar sensors and phased-array receivers. The receiver supports two channels which can be used to support sum and delta antenna pattern inputs. The receiver comprises of two LNAs, a DPST switch, an I/Q downconverter, baseband variable gain amplifiers, and integrate-and-dump filters. The receiver has 45 dB of conversion gain with 7.8 dB noise figure (with plastic package) at 24 GHz.

A set of calibration standards including multiple reflects is demonstrated for performing multimode thru-reflect-line (MTRL) calibrations. The measured multimode propagation characteristics for these lines are shown and the accuracy of the calibration using a particular reflect standard is demonstrated to 40 GHz. Small-signal multimode S-parameter measurements of a power amplifier demonstrate the value of the measurements for providing differential results including the common-mode rejection and mode-conversion parameters.

The design of a balanced, three-stage, common-emitter, 24 GHz SiGe HBT power amplifier for ultra-wideband applications is described. The unique features of the amplifier are very flat gain response in the frequency band of interest and sharp gain drop outside of the band, which are important considerations for a system-on-a-chip UWB application. The amplifier has 18 dB nominal gain in the frequency band of 24/spl plusmn/2 GHz. The gain variation is /spl plusmn/0.5 dB in the same frequency band. Saturated output power is 12 dBm at 24 GHz.

We describe an optimization method to approximate the arrival rate of data such as e-mail messages, website visits, changes to databases, and changes to websites mirrored by other servers. We model these arrival rates as non-homogeneous Poisson process based on observed arrival data. We estimate the arrival function by cubic splines using the maximum likelihood principle. A critical feature of the model is that the splines are constrained to be everywhere nonnegative. We formulate this constraint using a characterization of nonnegative polynomials by positive semidefinite matrices. We also describe versions of our model that allow for periodic arrival rate functions and input data of limited precision. We formulate the estimation problem as a convex program related to semidefinite programming and solve it with a standard nonlinear optimization package called KNITRO. We present numerical results using both an actual record of e-mail arrivals over a period of sixty weeks, and artificially generated data sets. We also present a cross-validation procedure for determining an appropriate number of spline knots to model a set of arrival observations.

Design of a low-loss high-isolation absorptive GaAs SPDT PIN diode switch at 24 GHz is presented. The switch incorporates a total of 6 PIN diodes, i.e., 3 diodes in each arm. Two of the diodes are for high isolation purposes and the third diode provides the matched load to the isolated port. The measured insertion loss and isolation are better than 1.4 dB and 47 dB, respectively, in the frequency band 22 to 26 GHz. The input and output return losses are better than 12 dB in the same frequency band

A novel wide-band DC-blocked MMIC RF termination with biasing network is presented. It provides return loss better than 18 dB from 100 MHz to 40 GHz. The main advantage of the design is that the DC blocking capacitor is placed directly on top of a wide grounding Si pedestal, thanks to M/A-COM's HMIC technology, reducing the ground inductance. At the RF side of the circuit, two 100-Ohm nichrome resistors have been put in parallel giving 50 Ohm. A small stub between the 100-Ohm resistors compensates for the inductance giving a very good match up to 40 GHz, Bond-wire pads have been designed on the chip and also on the host PCB to match the inductance of double wire-bond at the RF port.

A low-loss high-isolation GaAs SPST PIN diode switch for millimeter wave applications is presented. Modeling of PIN diode switches using full-wave electromagnetic simulators by including the semiconductor properties of the diode mesa layers is explained. This approach eliminates the necessity of model extraction from measurements in most cases. In addition, usage of simulation tools allows the optimization of diode structure for minimum insertion loss. Main loss mechanisms of GaAs PIN diodes at mm-wave frequencies are explained. Measurement results of a newly developed CPW SPST 55-85 GHz GaAs PIN switch are presented and compared with simulations. The developed shunt switch has 0.6 dB insertion loss and 20 dB return loss in the off state (through) at 75 GHz. It has 20 dB insertion loss and 1 dB return loss in the on state (isolation) at the same frequency.

An X-ray tensile testing technique for studying the stress transfer across polyimide passivation layers is reviewed. The advantage of this direct technique over indirect techniques is discussed, and results are presented for a model system consisting of a nickel substrate, a polyimide passivation layer, and a copper conducting layer. The stress transfer behavior is compared to the behavior predicted by theory, and the ability of the X-ray tensile technique to measure appropriate parameters for modeling the stresses in multilayer thin-film structures is discussed. In addition, the effect of polyimide thickness on stress transfer is reviewed, and the effect of copper line width on stress transfer is reported. The assumption of an ideal interface behavior is not valid for calculating the stress transfer across passivation layers in the loading geometry used for this study. The stress values in the Cu thin films were about 70% less than the predicted values. As predicted by the closed form solution, there is no effect of PI properties and thickness on the stress transfer behavior at room temperature. >

The data obtained from bulk test specimens are often used in deformation models to predict the mechanical responses of parts that are very different in size and shape, such as solder joints in VLSI surface mount components. Such predictions are valid only if the size of the solder joints exceeds the representative deformation volume of the bulk specimens. In the present study, the measurement of the representative volume was performed for a single-phase solder alloy (98Pb-2Sn wt.%) deformed in both tension and cyclic loading. Deformation of 98Pb-2Sn alloy at room temperature was shown to be extremely inhomogeneous in both tensile and cyclic loading, and strain localization was observed at the grain boundaries. Strain distribution is more inhomogeneous in the specimen tested in fatigue than the specimen tested in uniaxial tension. This may be due to fatigue inhomogeneity in strain distribution or a small degree of buckling. The number of grains required for a representative volume of the 98Pb-2Sn alloy specimens tested is quite large. >