In this paper, we will propose a new framework which can estimate the desired signal and the instrument response function (IRF) simultaneously from the degraded spectral signal. Firstly, the spectral signal is considered as a distribution, thus, new entropy (called differential-entropy, DE) is defined to measure the distribution with a uniform distribution, which allows negative value existing. Moreover, the IRF is parametrically modeled as a Lorentzian function. Comparative results manifest that the proposed method outperforms the conventional methods on peak narrowing and noise suppression. The deconvolution IR spectrum is more convenient for extracting the spectral feature and interpreting the unknown chemical mixtures.
A method for spectral estimation is proposed. It is based on the multidimensional extensions of the RELAX algorithm. The fast Fourier transform is replaced by multiple Chirp-Z transforms. Each transform has a much shorter length than the transform in the original algorithm. This reduces the memory requirements significantly. At the same time a high degree of parallelism is preserved. A detailed analysis of the computational requirements is given. Finally, the proposed method is applied to automotive radar measurements. It is shown, that the multidimensional spectral estimation resolves multiple scattering centers on an extended object.
In this paper, a performance limit is derived for a distributed Bayesian parameter estimation problem in sensor networks where the prior probability density function of the parameter is known. The sensor observations are assumed conditionally independent and identically distributed given the parameter to be estimated, and the sensors employ independent and identical quantizers. The performance limit is established in terms of the best possible asymptotic performance that a distributed estimation scheme can achieve for all possible sensor observation models. This performance limit is obtained by deriving the optimal probabilistic quantizer under the ideal setting, where the sensors observe the parameter directly without any noise or distortion. With a uniform prior, the derived Bayesian performance limit and the associated quantizer are the same as the previous developed performance limit and quantizers under the minimax framework, where the parameter is assumed to be fixed but unknown. This proposed performance limit under distributed Bayesian setting is compared against a widely used performance bound that is based on full-precision sensor observations. This comparison shows that the performance limit derived in this paper is comparatively much tighter in most meaningful signalto- noise ratio (SNR) regions. Moreover, unlike the unquantized observations performance limit which can never be achieved, this performance limit can be achieved under certain noise observation models.
In this paper, we investigate the effect of four different message schedules on the performance of an OFDM receiver with unknown carrier frequency and phase noise offsets. One of the methods uses a serial schedule and the remaining three techniques use non-serial message schedules. The serial schedule is shown to give good estimates in less number of iterations compared to the non-serial message schedules. The results also show that fast graphical estimators can be designed by using non-serial message schedules with damping to approach the bit error rates of the serial schedule while reducing the computational time for convergence. In particular, the damped flooding message schedule using four iterations reduces the computational time for convergence by more than 30 % compared to the serial message schedule using one iteration for signal to noise ratios lower than 20 dB.
We address the problem of extracting a time-varying fundamental frequency from a signal which has multiple, possibly aliased, harmonics, observed in potentially very high noise. The approach consists of an ML detector employing compressed likelihoods, followed by one of two processing stages which filter out unreasonable detections: either a target tracking approach or a Viterbi algorithm. Results show very good ability to extract the fundamental, even in very noisy data.
For many decades Digital Signal Processing (DSP) nodes have been designed for processing digital data received from arrays of radio telescopes. Common threads in all these nodes are: digital communications, processing and memory. Fundamentally the aim of each system was to provide the greatest operational capability for the technology available at that time. As the systems grew in size it became apparent that a key performance indicator was how processing nodes communicated. Poor communication could result in delayed schedules, reduced operational performance and higher system costs. The Square Kilometre Array (SKA) project represents a quantum leap in system size relative to current radio astronomy telescopes. This paper explores current work in this area and introduces the possibility of a fully optically connected processing and memory node. Such a node could be utilized for multi-stage polyphase filterbanks, beamforming and correlation. The application presented here is radio astronomy, but it could also be applied to defence and telecommunication systems.
Distributed detection with dependent observations is always a challenging problem. The problem of detection with shared information has many applications when sensors have overlapped measurements, e.g., when distributed detection is performed in a security system where sensors have overlapped coverages. For this shared information scenario, we investigate the distributed detection problem in parallel fusion networks. The design problem is how to best utilize the common information at both the local sensors and the fusion center to achieve best possible performance. We derive the necessary condition for the optimal sensor decision rules for all sensors. In addition, we investigate the system performance by comparing the optimal rules with suboptimal rules for distributed detection of a constant signal corrupted by Gaussian noise. The numerical results obtained by conducted examples confirm the optimality of the derived decision rules.
Under-determined mixtures in blind source separation (BSS) are characterized by the case that they have more inputs than outputs. The classical independent component analysis (ICA) methods cannot be applied to the under-determined case. However, sparseness-based approaches can be applied to the under-determined BSS. Two steps method has been widely employed to solve the under-determined BSS problem: mixing matrix estimation and source recovery. Source recovery in under-determined BSS (UBSS) is an NP -hard problem and, therefore, does not have a closed form solution. In this paper, we proposed a new blind non-negative source recovery approach to the under-determined mixtures. The results presented in this paper are limited to non-negative sources. Simulation results illustrate the effectiveness of our method.
This paper describes an acoustical investigation on Thai speech segmentation using a combination of average level crossing rate (ALCR) and root-mean-square (RMS) energy. Simple and easy to compute, ALCR information alone was successfully used in an automatic speech segmentation system for English. However, ALCR has never been applied to Thai. As a result, the objective of the study is to apply ALCR information to ascertain its usefulness in detecting significant temporal changes in continuous Thai Speech. An experiment was conducted on a small speech corpus containing 21 sentences. Preliminary results suggest that ALCR and RMS energy can be used to detect the phonetic boundary between obstruent initial consonant and preceding/following vowel. In addition, it can also be used to detect boundary between final consonant of the preceding syllable and initial consonant of the following syllable except for the case involving two successive non-identical nasals. The overall accuracy is around 81% for data from four speakers.
Palmprint is one of the most useful physiological biometrics that can be used as a powerful means in personal recognition systems. The major features of the palmprints are palm lines, wrinkles and ridges, and many approaches use them in different ways towards solving the palmprint recognition problem. Here we have proposed to use a set of statistical and wavelet-based features; statistical to capture the general characteristics of palmprints; and wavelet-based to find those information not evident in the spatial domain. Also we use two different classification approaches, minimum distance classifier scheme and weighted majority voting algorithm, to perform palmprint matching. The proposed method is tested on a well-known palmprint dataset of 6000 samples and has shown an impressive accuracy rate of 99.65%-100% for most scenarios.
In this paper, we developed a mathematical model for a single-hop relay-based communication channel. Assuming the transmitter-to-relay and receiver-to-relay channels are non-line-of-sight flat fading channels, we show that the real and imaginary components of the combined single channel have Laplace probability density functions. We, therefore, develop a complex Laplace autoregressive process (AR) that captures the statistical characteristics of the fading process of the relay channel. Such an AR model makes channel simulations simple, and eases formulation of such problems as data detection in a state-space form for convenient application of well-known algorithms. Furthermore, the autocorrelation of the developed Laplace AR model has Yule-Walker type of properties that enables us to configure its parameters to match to the second-order statistical characteristics of the channel through autocorrelation matching. The derivation of the channel model is illustrated through an example and computer simulations.
Cryo-electron tomography (cryo-ET), which produces three dimensional images at molecular resolution, is one of many applications that requires image reconstruction from projection measurements acquired with irregular measurement geometry. Although Fourier transform based reconstruction methods have been widely and successfully used in medical imaging for over 25 years, assumptions of regular measurement geometry and a band limited source cause direction sensitive artifacts when applied to cryo-ET. Iterative space domain methods such as compressed sensing could be applied to this severely underdetermined system with a limited range of projection angles and projection length, but progress has been hindered by the computational and storage requirements of the very large projection matrix of observation partials. In this paper we derive a method of dynamically computing the elements of the projection matrix accurately for continuous basis functions of limited extent with arbitrary beam width. Storage requirements are reduced by a factor of order 107 and there is no access overhead. This approach for limited angle and limited view measurement geometries is posed to enable dramatically improved reconstruction performance and is easily adapted to parallel computing architectures.
Array radio telescopes are suitable for the implementation of spatial filters. These filters present the advantage of canceling potential radio frequency interference (RFI) while recovering uncorrupted Time-Frequency data, of interest to astronomers. Although information regarding the sources of RFI can be a priori known or reliably inferred, the complexity of radio telescope systems randomizes the formulation of the subspace spanned by the RFI due to a lack of calibration or characterization. This knowledge is however necessary for building an efficient spatial filter, and needs therefore to be estimated.
This paper describes our efforts to include a hands-on component in the teaching of core concepts of digital signal processing. The basis of our approach was the low-cost and open-source "Stanford Lab in a Box." This system, with its easy to use Arduino-like programming interface allowed students to see how fundamental DSP concepts such as digital filters, FFT, and multi-rate processing can be implemented in real time on a fixed-point processor. The paper describes how the Lab in a Box was used to provide a new dimension to the teaching of DSP.
Software defined radio (SDR) is an exciting merger of digital signal processing and wideband radio hardware. The term SDR came into more common usage in 1992 by Dr. Joe Mitola, but actually had its beginnings back in 1984 at E-Systems. The ideal SDR receiver consists of an antenna connected to an analog-to-digital converter (ADC) followed by a digital signal processing system (DSPS) to extract the signal of interest. Low-cost, as in $20, SDR receivers originally designed for digital video broadcasting, have been available for several years. Giving undergraduate students hands-on experience in this area is needed. In this paper we describe the details of an SDR laboratory experiment for students in a first semester communications theory course. Complete open-source SDR receiver software is used to get started, then coding of DSP algorithms is explored to process captured radio signals generated using test equipment and then actual over-the-air broadcasts. Being able to write code to process live signals and then see and hear the results really connects with the students. Both Matlab and Python support code libraries are available.
Based on the compressive sensing (CS) theory, it is possible to recover signals, which are either compressible or sparse under some suitable basis, via a small number of non-adaptive linear measurements. In this paper, we investigate recovering of block-sparse signals via multiple measurement vectors (MMVs) in the presence of noise. In this case, we consider one of the existing algorithms which provides a satisfactory estimate in terms of minimum meansquared error but a non-sparse solution. Here, the algorithm is first modified to result in sparse solutions. Then, further modification is performed to account for the unknown block sparsity structure in the solution, as well. The performance of the proposed algorithm is demonstrated by experimental simulations and comparisons with some other algorithms for the sparse recovery problem.
Detecting and localizing a person crossing a line segment, i.e., border, is valuable information in security systems and human context awareness. To that end, we propose a border crossing localization system that uses the changes in measured received signal strength (RSS) on links between transceivers deployed linearly along the border. Any single link has a low signal-to-noise ratio because its RSS also varies due to environmental change, (e.g., branches swaying in wind), and sometimes does not change significantly when a person crosses it. The redundant, overlapping nature of the links between many possible pairs of nodes in the network provides an opportunity to mitigate errors. We propose new classifiers to use the redundancy to estimate where a person crosses the border. Specifically, the solution of these classifiers indicates which pair of neighboring nodes the person crosses between. We demonstrate that in many cases, these classifiers provide more robust border crossing localization compared to a classifier that excludes these noisy, redundant measurements.
In source localization applications, coherency among the signals is an important source of error for parameter estimation. In this paper, a method is proposed to solve the localization problem where there are coherently mixed arbitrary number of far- and near-field sources. In order to estimate the direction-of-arrival (DOA) and the range parameters, compressed sensing (CS) approach is presented where a dictionary matrix is constructed with far- and near-field steering vectors. A sparse vector including the supports of the source signals is estimated in spatial domain. The supports of coherent signals are recovered by using convex minimization techniques. It is shown that the proposed approach recovers the signal components of the array output as well as determining the source locations.
The ASKAP radio telescope in Australia is the first synthesis imaging array to use phased-array feeds (PAFs). These permit wider fields of view and new modalities for radio-frequency interference (RFI) mitigation. Previous work on imaging-array RFI cancellation has assumed that processing bandwidths are very narrow, and correlator integration times are short. However, these assumptions do not necessarily reflect real-world instrument limitations. This paper explores adaptive array cancellation algorithm effectiveness on ASKAP for realistic bandwidths and integration times. With ASKAP's beamforming PAFs on each dish, followed by a central correlation processor across beamformed signals from all dishes, one may consider algorithms that span multiple levels in the hierarchical signal processing chain. We compare performance for several subspace-projection-based algorithms applied to different tiers of this extended architecture. Simulation results demonstrate that it is most effective to cancel at the PAF beamformers.
The Stanford "Lab-In-A-Box" project comprises an open source hardware and software tool chain for teaching signal processing and analog electronics. It is intended to improve the teaching of these concepts by providing a platform that is more open and understandable and by lowering the economic barriers to students interested in the field. To do this, the Lab-In-A-Box brings a full powered Digital Signal Processor (DSP) core to the popular Arduino microcontroller environment and marries it with a simple to use analog front end (AFE). The software platform provided with the Lab-In-A-Box includes an Arduino-like development environment that facilitates learning and quick development of signal processing applications without abstracting away the intricacies of a practical implementation. This system has been used to create several teaching examples and has been tested in courses at Stanford University.
