Analysis of integrated optical waveguide mirrors

A new method is presented to analyze reflection losses of integrated mirrors, taking into account the exact guided mode profile and assuming that this profile remains unchanged up to the reflecting plane. The fraction of the reflected light coupled to one of the guided modes of the output waveguide is calculated, taking into account the mirror reflection coefficient. The influence of both translation and tilt of the reflecting plane is investigated. The method applies for every guided mode and any reflection angles. Numerical calculations are derived for a 90° optical corner mirror     

Error floor of MSK modulation in a mobile-radio channel with twoindependently fading paths

We investigate the error probability bit error rate (BER) of minimum shift keying (MSK) modulation with differential detection in a two-path fading channel without noise (error floor). We develop a new method for the computation of the BER: we show that errors occur if the phasors of the instantaneous impulse response fall into certain regions of the complex plane; then we average over the statistics of the phasors to arrive at the mean BER. With this method, we derive analytical expressions for the BER for arbitrary amplitude statistics of the paths. For the special case of two Rayleigh-fading paths with small delay, we find that the BER is proportional to the square of the mean delay spread (normalized to the bit length) if we sample between the two pulses. This proves the qualitative behavior of previous estimates, but our results allow also a more exact quantitative formulation. The quadratic dependence of the BER on the delay spread breaks down if we have one Rayleigh-fading and one Rician-fading path. We find that the bit combinations 1-11 and -11-1 do not lead to errors in the two-path model. However, additional Monte Carlo simulations show that these bit combinations do lead to errors in a three-path model     

Design, manufacture and evaluation of a scalable high-performanceneural system

The authors describe a scalable neural system, HyperNet, based on a probabilistic RAM-based architecture and using a custom VLSI IC. A system using five HyperNet VLSI ICs and capable of realising up to 10240 neurons has been designed, manufactured and demonstrated to have the potential to learn more than three orders of magnitude faster than simulations on current workstations     

ACT3: a high-speed, high-precision electrical impedance tomograph

Presents the design, implementation, and performance of Rensselaer's third-generation adaptive current tomograph, ACT3. This system uses 32 current sources and 32 phase-sensitive voltmeters to make a 32-electrode system that is capable of applying arbitrary spatial patterns of current. The instrumentation provides 16 b precision on both the current values and the real and reactive voltage readings and can collect the data for a single image in 133 ms. Additionally, the instrument is able to automatically calibrate its voltmeters and current sources and adjust the current source output impedance under computer control. The major system components are discussed in detail and performance results are given. Images obtained using stationary agar targets and a moving pendulum in a phantom as well as in vivo resistivity profiles showing human respiration are shown     

Measurement and modeling of the sidewall threshold voltage ofmesa-isolated SOI MOSFETs

Five-terminal silicon-on-insulator (SOI) MOSFETs have been characterized to determine the threshold voltage at the front, back, and sidewall as a function of the body bias. The threshold voltage shift with the body bias at the front and back interfaces can be explained by the standard bulk body effect equation. However, the threshold voltage shift at the sidewall is smaller than predicted by this equation and saturates at large body biases. This anomalous behavior is explained by two-dimensional charge sharing between the sidewall and the front and back interfaces. An analytical model that accounts for this charge sharing by a simple trapezoidal approximation of the depletion regions and correctly predicts the sidewall threshold voltage shift and its saturation is discussed. The model makes it possible to measure the sidewall threshold even when it is larger than the front threshold voltage     

On the existence and the stability of solutions for higher-order semilinear Dirichlet problems

We investigate the existence and stability of solutions for higher-order two-point boundary value problems in case the differential operator is not necessarily positive definite, i.e. with superlinear nonlinearities. We write an abstract realization of the Dirichlet problem and provide abstract existence and stability results which are further applied to concrete problems.     

Shape transition from InAs quantum dash to quantum dot on InP(3 1 1)A

We report on the shape transition from InAs quantum dashes to quantum dots (QDs) on lattice-matched GaInAsP on InP(3 1 1)A substrates. InAs quantum dashes develop during chemical-beam epitaxy of 3.2 monolayers InAs, which transform into round InAs QDs by introducing a growth interruption without arsenic flux after InAs deposition. The shape transition is solely attributed to surface properties, i.e., increase of the surface energy and symmetry under arsenic deficient conditions. The round QD shape is maintained during subsequent GaInAsP overgrowth because the reversed shape transition from dot to dash is kinetically hindered by the decreased ad-atom diffusion under arsenic flux.