Signal Processing Community Research Articles

Recent development of speech and audio signal processing in network communication

Speech and audio signal processing plays an important role in human interactive communication over networks. With the development of wide-band and mobile networks, humans are w illing to exchange more fruitful and fluent speech and audio contents over the networks. These developments advance the techniques of speech and audio signal processing including systems, theories, algorithms and processing tools.

Signal Processing for Gigabit-Rate Wireline Communications: An Overview of the State of the Art and Research Challenges

Signal processing played an important role in improving the quality of communications over copper cables in earlier digital subscriber line (DSL) technologies. Even more powerful signal processing techniques are required to enable a gigabit per second data rate in the upcoming fast access to subscriber terminals (G.fast.) standard. This new standard is different from its predecessors in many respects. In particular, G.fast will use a significantly higher bandwidth. At such a high bandwidth, crosstalk among different lines in a binder will reach unprecedented levels, which are beyond the capabilities of most efficient techniques for interference mitigation.

Unambiguous Techniques Modernized GNSS Signals: Surveying the solutions

Global navigation satellite systems (GNSSs) remain, by far, the most reliable and widely spread sources of accurate outdoor positioning. With the advent of the new and modernized signals, novel challenges related to GNSS receivers have also been recognized. One of the most important challenges in dealing with modernized GNSS signals is posed by the oscillatory behavior of the autocorrelation of some of those signals, which causes ambiguities in the measurement of the propagation delay. This article is an extensive survey of the solutions in this area proposed during past decades. Not only does it illustrate that a large pool of solutions is available, but it also shows that none of these solutions alone can currently overcome all of the challenges related to GNSS ambiguities. A thorough overview of the problems caused by the ambiguities in the delay estimation and the unambiguous techniques proposed to counteract them is presented. We hope to spark the interest of the signal processing community and to stimulate new advances in this field. Unambiguous methods are classified into three main classes, and we compare the main solutions in terms of complexity and performance to identify the most promising techniques and directions to be followed. We point out that there is an inherent tradeoff between the unambiguous acquisition and unambiguous tracking, and that the receiver stages of acquisition and tracking can be designed in a disjoint manner when dealing with the ambiguities.

Communication Models for Distributed Acoustic Sensing for Telemetry

Distributed acoustic sensing (DAS) works as an efficient tool for monitoring acoustic signals in adverse environments, such as those commonly found in geophysical explorations, oil & gas, and military applications. DAS employs fiber-optic cables in order to record the acoustic wave impinging on them. These sensors can be several kilometers long, and its distributed capacity allows the acoustic wave to be measured with an adjustable space-frequency resolution. Recent DAS advances have shown that this technology is a formidable tool for implementing communication/telemetry systems in hostile environments. Notwithstanding the increasing practical interest in this technology, communication and signal processing communities have not concentrated proper efforts on addressing the myriad of challenges this technology raises. This paper bridges this gap by describing the basic signal model of a DAS system in a style compatible with the common background of these communities. It first derives, in an ideal setup, the relation between pressure signal and acquired baseband electric signals. The pressure signal is assumed to be a consequence of acoustic activity, whereas, the electric signals are obtained through the use of photodiodes which sense pressure-modulated light signals. The consequences of using long-pulse signals are also studied considering a simple continuous model for the reflections at different spatial regions of the fiber, called Rayleigh backscattering phenomenon. This paper includes results that compare the theoretical proposals with practical signals acquired in controlled experiments.

Introduction to the Issue on Signal Processing and Machine Learning

The papers in this special section focus on the use of machine learning and signal processing for educational applications. Recent advances in machine learning and signal processing provide promising new avenues to move beyond this “one size fits all” educational approach. The key is that today’s learning technology platforms can capture personal learning data about students as they proceed through courses: performance on homework and exams, click actions made while watching lecture videos or interacting with simulations, the social learning networks formed among the students, the content posted on discussion forums, and so on. This data enables new opportunities to study the process of student learning and to design systems that improve learning at scale by closing the learning feedback loop. This special issue of showcases research from the signal processing community that is advancing effective learning at scale.

Nonlinear processing of a multicomponent communication signal by combination-sensitive neurons in the anuran inferior colliculus.

Diverse animals communicate using multicomponent signals. How a receiver's central nervous system integrates multiple signal components remains largely unknown. We investigated how female green treefrogs (Hyla cinerea) integrate the multiple spectral components present in male advertisement calls. Typical calls have a bimodal spectrum consisting of formant-like low-frequency (~0.9kHz) and high-frequency (~2.7kHz) components that are transduced by different sensory organs in the inner ear. In behavioral experiments, only bimodal calls reliably elicited phonotaxis in no-choice tests, and they were selectively chosen over unimodal calls in two-alternative choice tests. Single neurons in the inferior colliculus of awake, passively listening subjects were classified as combination-insensitive units (27.9%) or combination-sensitive units (72.1%) based on patterns of relative responses to the same bimodal and unimodal calls. Combination-insensitive units responded similarly to the bimodal call and one or both unimodal calls. In contrast, combination-sensitive units exhibited both linear responses (i.e., linear summation) and, more commonly, nonlinear responses (e.g., facilitation, compressive summation, or suppression) to the spectral combination in the bimodal call. These results are consistent with the hypothesis that nonlinearities play potentially critical roles in spectral integration and in the neural processing of multicomponent communication signals.

What Do You Consider a Successful Career? : Perspectives from signal processing-trained professionals [Community Voices

The motivation behind this column is to strengthen ties with readers and members in the signal processing community. In doing so, we set out to collect reflections from diverse members of our community on questions that are of interest to many. Here, examines the training necessary for a career in signal processing and explores how one can determine success in one's career in this industry.

An efficient privacy-preserving scheme for secure network coding based on compressed sensing

Network coding (NC) provides an elegant solution for improving capacity and robustness in computer networks. Different to traditional “store-and-forward” transmission paradigm, each intermediate node linearly combines received data packets, and the original files can be decoded at the sink nodes in NC settings. This brand-new paradigm is vulnerable to pollution attack, which means that some malicious nodes inject fake data packets into the network and this will lead to incorrect decoding. There are some information-theoretical solutions and cryptographic solutions for solving this security issue, and most existing schemes can thwart data pollution attacks. However, the privacy of the original files are vital to some application environments (e.g. military network). To the best of our knowledge, there is not a secure scheme which can thwart pollution attack and can protect the privacy of transmitted data simultaneously. In this paper, we present an efficient privacy-preserving scheme for secure network coding based on compressed sensing (CS), which has attracted considerable research interest in the signal processing community. Specifically, we embed CS into the general NC framework, i.e., the source node needs to compress each original data packet using the sensing matrix before creating the augmented vector and the sink nodes require to perform an additional CS reconstruction algorithm for reconstructing the original file. In addition, we construct a simple key distribution protocol and each intermediate node just needs two secret keys for verifying the integrity of received data packets. Such novel hybrid construction enables the privacy-preserving guarantee, and the performance comparison shows the high-efficiency of our scheme in terms of the computational complexity and communication overhead.

Signal Processing for Social Good [In the Spotlight

Communication, speech processing, seismology, and radar are well known applications of signal processing that contribute to the betterment of humanity. But is there a more direct way that signal and information processing can reduce poverty, hunger, inequality, injustice, ill health, and other causes of human suffering? The member states of the United Nations ratified 17 sustainable development goals in 2015, which, if achieved by the targeted year 2030, will end or greatly curtail these problems. Achieving the global goals, however, will require cooperation from all, including the signal processing community.

An Overview on Time/Frequency Modulated Array Processing

Time and frequency modulated arrays have numerous application areas including radar, navigation, and communications. Specifically, a time modulated array can create a beampattern with low sidelobes via connecting and disconnecting the antenna elements from the feed network, while the frequency modulated frequency diverse array produces a range-dependent pattern. In this paper, we aim to introduce these advanced arrays to the signal processing community so that more investigations in terms of theory, methods, and applications, can be facilitated. The research progress of time/frequency modulated array studies is reviewed and the most recent advances are discussed. Moreover, potential applications in radar and communications are presented, along with their technical challenges, especially in signal processing aspects.

A Novel multiple kernel-based dictionary learning for distributive and collective sparse representation based classifiers

In recent years, sparse representation theory has attracted the attention of many researchers in the signal processing, pattern recognition and computer vision communities. The choice of dictionary matrix plays a key role in the sparse representation based methods. It can be a pre-defined dictionary or can be learned via an optimization procedure. Furthermore, the dictionary learning process can be extended to a non-linear setting using an appropriate kernel function in order to handle non-linear structured data. In this framework, the choice of kernel function is also a key step. Multiple kernel learning is an appealing strategy for dealing with this problem. In this paper, within the framework of kernel sparse representation based classification, we propose an iterative algorithm for coincident learning of the dictionary matrix and multiple kernel function. The weighted sum of a set of basis functions is considered as the multiple kernel function where the weights are optimized such that the reconstruction error of the sparse coded data is minimized. In our proposed algorithm, the sparse coding, dictionary learning and multiple kernel learning processes are performed in three steps. The optimization process is performed considering two different structures namely distributive and collective for the sparse representation based classifier. Our experimental results show that the proposed algorithm outperforms the other existing sparse coding based approaches. These results also confirm that the collective setting leads to better results when the number of training examples is limited. On the other hand, the distributive setting is more appropriate when there are enough training samples.

Wideband delay and Direction of Arrival Estimation using sub-Nyquist sampling

Estimating the Direction of Arrival (DOA) of the signals received by a sensor array has been at the core of the array signal processing community for decades. With the advancement of the technology, the limits on the data and data transfer are increasing day by day. Such demands of the increasing bandwidths put an immense burden on the Analog to Digital Converters (ADCs). On the other hand, in the literature most of the wideband DOA estimation algorithms depend on the narrowband signal model where the signal is first decomposed into multiple signals with narrow frequency bins, the DOA is estimated for all of them and the outputs are combined together (coherent or non-coherent) to get the final result. This process is highly computational and highly susceptible to any outliers. Compressed Sensing (CS) over the last few years has shown a great promise in overcoming these problems while preserving the structure of the signal provided it is sparse in some domain. In this paper, we propose to estimate the DOA of wideband signals using CS based reconstruction recognizing that the autocorrelation of such signals is sparse in nature. The peaks in the correlation plot represent the time delays of the corresponding signals which is directly related to their DOAs. We identify and show that the relation between the cross-correlations of the sub-Nyquist signal and the Nyquist rate signal is a block sparse CS problem. By using a block sparse CS reconstruction algorithm we are able to estimate the peak locations in the Nyquist correlation vector thereby estimating the DOAs of the wideband signals.

Adaptive and Robust Sparse Coding for Laser Range Data Denoising and Inpainting

Sparse coding (SC) is making a significant impact in computer vision and signal processing communities, which achieves the state-of-the-art performance in a variety of applications for images, e.g., denoising, restoration, and synthesis. We propose an adaptive and robust SC algorithm exploiting the characteristics of typical laser range data and the availability of both range and reflectance data to realize range data denoising and inpainting. Specifically, our method estimates the informative level of each patch according to the variation in both range and reflectance modalities, followed by adaptive dictionary training that assigns dynamic sparsity weights to the patches with different informative levels. Furthermore, the $\ell _{1}$ -norm-based representation fidelity measure is applied to make our method robust to outliers which are common in laser range measurements. Extensive experiments on synthetic and real data demonstrate that our method works effectively, resulting in superior performance both visually and quantitatively, compared with competitive methods including the available sparse-representation-based algorithm, wavelets, partial differential equation, and non-local means.

Fundamental communications theories and signal processing techniques for amorphous cellular systems [Guest Editorial

The rapid developing of the fourth generation (4G) wireless communications has aroused tremendous demands for high speed data transmission due to the dissemination of various types of the intelligent user terminals as well as the wireless multi-media services. It is predicted that the network throughput will increase 1000 times till year 2020 as compared to that of year 2010.

Smart Driver Monitoring: When Signal Processing Meets Human Factors: In the driver's seat

This article provides an interdisciplinary perspective on driver monitoring systems by discussing state-of-the-art signal processing solutions in the context of road safety issues identified in human factors research. Recently, the human factors community has made significant progress in understanding driver behaviors and assessed the efficacy of various interventions for unsafe driving practices. In parallel, the signal processing community has had significant advancements in developing signal acquisition and processing methods for driver monitoring systems. This article aims to bridge these efforts and help initiate new collaborations across the two fields. Toward this end, we discuss how vehicle measures, facial/body expressions, and physiological signals can assist in improving driving safety through adaptive interactions with the driver, based on the driver's state and driving environment. Moreover, by highlighting the current human factors research in road safety, we provide insights for building feedback and mitigation technologies, which can act both in real time and postdrive. We provide insights into areas with great potential to improve driver monitoring systems, which have not yet been extensively studied in the literature, such as affect recognition and data fusion. Finally, a high-level discussion is given on the challenges and possible future directions for driver monitoring systems.

Modelling hospital length of stay using convolutive mixtures distributions.

Length of hospital stay (LOS) is an important indicator of the hospital activity and management of health care. The skewness in the distribution of LOS poses problems in statistical modelling because it fails to adequately follow the usual traditional distribution of positive variables such as the log-normal distribution. We present in this paper a model using the convolution of two distributions, a technique well known in the signal processing community. The specificity of that model is that the variable of interest is considered to be the resulting sum of two random variables with different distributions. One of the variables features the patient-related factors in terms of their need to recover from their admission condition, while the other models the hospital management process such as the discharging process. Two estimation procedures are proposed. One is the classical maximum likelihood, while the other relates to the expectation-maximization algorithm. We present some results obtained by applying this model to a set of real data from a group of hospitals in Victoria (Australia). Copyright © 2016 John Wiley & Sons, Ltd.

Star-type polarizer with equal-power splitting function for each polarization based on polarization-dependent defects in two-dimensional photonic-crystal waveguides.

We propose a star-type polarizer with equal-power splitting function for each polarization based on polarization-dependent defects (PDDs) in two-dimensional photonic-crystal waveguides (PCWs). The structure is designed by combining two Y-type PCWs, and two types of PDDs are introduced into the PCWs respectively to provide polarization functions. By using finite-element method and optimizing the parameters of the PDDs, it is demonstrated that different polarizations can only transmit through their own PCWs and output with identical power distributions, i.e., the structure can function as polarizer and equal-power splitter for each polarization at the same time. In addition, by scanning the wavelength of the structure, it is proved that the proposed splitter can work in a wide range of wavelength while keeping high output transmission for both the TE and TM polarizations. Such a structure is useful for polarization-relative multi-channel signal processing for optical communications in the mid- and far-infrared wavelength regions.

Detection and communication in the “Beaufort Lens”

The WHOI Ice Tethered Profilers in the Beaufort has confirmed a “sound speed duct” at depths between 100 to 250 m labeled as the “Beaufort Lens.” It is thought to be caused by warm water intrusion from the Bering Strait. A significant consequence of the duct is the prediction of 10 dB lower transmission losses at ranges of 100 km for sources and receivers both within the duct. We examine how the properties of the duct impact detection and communication. Transmission loss is strongly affected by frequency. Ducted propagation is not supported below a modal cutoff and absorption losses become consequential at higher frequencies. For active sonar, where wideband waveforms are employed, dispersion effects and multipath become important. The shape of the duct, especially boundary gradients, impacts this dispersion. While a vertical line array spanning the duct can potentially exploit the multipath ray/mode coherence and the resolution of the array impacts whether coherent or incoherent combination should be employed to achieve desired gains. Data for validating and verifying long range transmission loss in the duct were acquired by two submarines participating in ICEX-16. This presentation examines the signal processing for detection and communication within the so called “Beaufort Lens.”

Noise-Loop Multiple Access

A fundamental problem for signal processing in wireless communications is the separation of multiple signals/users from their mixture. The problem is simpler if they are nonoverlapping in either the time domain or the frequency domain. Two signals may be overlapping in both time and frequency, but they can still be easily separated if they are uncorrelated (orthogonal), as in code-division multiple-access (CDMA) systems. It may be interesting to know that two signals may be overlapping in both the frequency and time domains and not orthogonal to each other, but they can still be separated if each autocorrelation function at the receiver is separable in the time-lag domain. In this paper, we will show how noise-loop (NL) modulation, which was originally thought to produce security at the physical layer, can be used to effectively separate uncoordinated/nonorthogonal users sharing the same time and frequency interval. To show the potential of the NL multiple-access (NLMA) scheme, the theoretical analysis is carried out and compared with simulations. In particular, together with the error probability, a new figure of merit is derived: the multiuser efficiency index. It can be shown that the NLMA scheme is able to distinguish the bit stream of the users, regardless of their number. In addition, the capacity region of the NLMA scheme is derived and compared with the conventional CDMA scheme. All of the aforementioned features, together with the intrinsic security, make the NL an interesting scheme for wireless technology.

Data-Driven Forward–Backward Pursuit for Sparse Signal Reconstruction

In recent years, compressed sensing has received considerable attention from the signal processing community because of its ability to represent sparse signals with a number of samples much less than that is required by the Nyquist sampling theorem. $$\ell _{1}$$l1-minimization is a powerful tool for sparse signal reconstruction from few measured samples, but its computational complexity is a burden for real applications. Recently, a number of greedy algorithms based on orthogonal matching pursuit (OMP) have been proposed, and they have near $$\ell _{1}$$l1-minimization performance with much less processing time. In this work, a new OMP-type two-stage sparse signal reconstruction algorithm, namely data-driven forward---backward pursuit (DD-FBP), is proposed. It is based on a former work called forward---backward pursuit (FBP). DD-FBP iteratively expands and shrinks the estimated support set, and these constitute the forward and backward stages. In DD-FBP, unlike FBP, the forward and backward step sizes are not constants, but they are dependent on the correlation and projection values, respectively, which are calculated in each iteration. The recovery performance by using noiseless and noisy sparse signal ensembles, as well as a natural sparse image, indicates that DD-FBP surpasses the other methods in terms of success rate and processing time.