Bernard C. Levy

We develop a resource-efficient framework for collaborative decision-making over distributed sensor networks by proposing a novel over-the-air soft information aggregation. We exploit the natural superposition of wireless transmissions to enable sensors to utilize over-the-air computation to approximate the sufficient statistic for optimum detection over a shared channel. By designing practical transmission and receiver processing in over-the-air computation, the decision-making fusion center can wirelessly obtain a good approximation of the aggregate log-likelihood ratio computed over all observed data with low distortion. Focusing on Neyman-Pearson tests for detection in this new framework, we develop efficient tests and analyze their performance bounds in several common joint detection scenarios. Our results show significant over-the-air collaboration gain even with a few participating sensors. The novel framework exhibits very little performance loss of detection accuracy against traditional multiple access transmission from sensing nodes despite substantial resource savings via over-the-air computation.

Classical receiver architectures demodulate a bandpass signal to baseband before sampling the in-phase and quadrature components. With the advent of faster analog-todigital converters (ADCs) and wide bandwidth sample and hold (S/H) circuits, it has become practicable to sample a bandpass signal directly without any demodulation operation and then process it with robust DSP technology. Direct sampling methods do present their own challenges. When a single channel is used to digitize the signal, not all frequencies above the Nyquist rate are allowed. As first shown by Kohlenberg, this restriction can be removed with a two-channel time-interleaved ADC (TIADC) where two ADCs separated by a timing offset independently sample the signal. In this paper, we propose a general and flexible technique for sampling the complex envelope of a bandpass signal using a nonuniformly interleaved M -channel TIADC. The signal is sampled directly by the sub-ADCs which are separated by timing skews. Reconstruction of the envelope entails inverting a matrix of filters resulting in some timing skews being forbidden. The proposed sampling scheme requires the implementation of M digital FIR filters and can be used to digitize bandpass signals with any carrier frequency in software defined radio applications. Reconstruction analysis is provided for the case of two, three, and four channels. Multi-tone and MSK signals are used in simulations to validate the proposed method and assess its performance.

A general and flexible technique is proposed for sampling the complex envelope of a bandpass signal by using a nonuniformly interleaved two-channel analog-to-digital converter (TIADC). The signal is sampled directly without any demodulation operation, but the two sampling channels are not uniformly interleaved, since some TIADC timing offsets are rendered forbidden. The proposed complex envelope sampling scheme requires the implementation of two digital FIR filters and a discrete-time modulator. Quadrature sampling, a special case that assumes certain parameter relations, is investigated as an approximation technique. Computer simulations are presented and illustrate a -57dB mean-square error (MSE) can be achieved via quadrature approximation for sufficiently large filter orders.

This paper describes a prototype two-step ADC with adjustable resolution in 40 nm CMOS technology. It uses a front-end successive-approximation-register ADC and a back-end time-domain ADC. Digital statistical calibration overcomes errors from front-end mismatch as well as inter-stage and back-end nonlinearity by analyzing specific statistical properties. In the 12-bit mode at 20 MS/s, the maximum SNDR is 59 dB before calibration and 68 dB after calibration, using 6.2 fJ per conversion-step, excluding the power dissipation required by the calibration.

Flavohemoglobins (FlavoHb) are metalloenzymes catalyzing the reaction of nitric oxide dioxygenation. The iron cation of the heme group needs to be preliminarily reduced to the ferrous state to be catalytically competent. This reduction is triggered by a flavin adenine dinucleotide (FAD) prosthetic group which is localized in a distinct domain of the protein. In this paper we obtain new insights into the internal long range electron transfer (over ca. 12 Å) using a combination of experimental and computational approaches. Employing a time-resolved pulse radiolysis technique we report the first direct measurement of the FADH˙ → HemeFe^III electron transfer rate. A rate constant of (6.8 ± 0.5) × 10³ s⁻¹ is found. A large panel of computational approaches are used to provide the first estimation of the thermodynamic characteristics of the internal electron transfer step within flavoHb: both the driving force and the reorganization energy are estimated as a function of the protonated state of the flavin semi-quinone. We also report an analysis of the electron pathways involved in the tunnelling of the electron through the aqueous interface between the globin and the flavin domains.

We propose a blind maximum-likelihood (ML) detection algorithm for continuous-phase modulated (CPM) signals transmitted over a noisy linear FIR channel. This is referred to as the Expectation-Maximization-Viterbi algorithm (EMVA) [1, 2]. The EMVA is a blind algorithm capable of simultaneously performing system identification and signal estimation whenever the transmission system can be modeled as a finite-state machine with unknown parameters, a scenario frequently encountered in signal processing for communications. We specifically focus on CPM, but t he algorithm is equally applicable to any other modulation type, linear or nonlinear. The channel estimate obtained via the EMVA is shown via simulations to, asymptotically (as the SNR increases) attain the most optimistic Cramér-Rao bo und and to have an error performance close to that of the true channel; a 0.5 dB difference is seen at BER equal to 10 ⁻⁴ .