An important source of uncertainty in tsunami forecasting arises from uncertainty in the event’s initial conditions. In this work, we propose a dictionary-based inversion method that uses off-shore sensor data to recover the initial ocean condition, allowing inversion of any tsunami event, and leading to reduced uncertainty in forecasts. We show that deep learning models can be used to address the computational requirements that arise from dictionary-based inversion. We validate our method using simulations of historic events.

For many inverse problems, the data on which the solution is based is acquired sequentially. We present an approach to the solution of such inverse problems where a sensor can be directed (or otherwise reconfigured on the fly) to acquire a particular measurement. An example problem is magnetic resonance image reconstruction. We use an estimate of mutual information derived from an empirical conditional distribution provided by a generative model to guide our measurement acquisition given measurements acquired so far. The conditionally generated data is a set of samples which are representative of the plausible solutions that satisfy the acquired measurements. We present experiments on toy and real world data sets. We focus on image data but we demonstrate that the method is applicable to a broader class of problems. We also show how a learned model such as a deep neural network can be leveraged to allow generalisation to unseen data. Our informed adaptive sensing method outperforms random sampling, variance based sampling, sparsity based methods, and compressed sensing.

The traditional vector median filtering and its variants used to reduce impulse noise in digital color images operate by processing over all the pixels in the image sequentially. This renders these filtering methods computationally expensive. This paper presents a fast method for reducing impulse noise in digital color images. The key idea here is to slice each row of the image as a univariate data vector, identify impulse noise using anomaly detection schemes and then apply median filtering over these to restore the original image. This idea ensures fast filtering as only the noisy pixels are processed. Using simulations, we show that the proposed method scales efficiently with respect to accuracy and time. Through a combined measure of time and accuracy, we show that the proposed method exhibits nearly 42% improvement over the conventional ones.

It can be challenging for opera singers to access their performance venues for rehearsal due to venue sched-ules or travel restrictions. Virtual reality (VR) or augmented reality (AR) technologies provide the possibility for rehearsal in a virtual venue. However, these technologies are mainly focused on visuals, rather than on the sonic plausibility of a virtual space. Moreover, existing self-auralization methods have one or more of the following limitations: a small choice of virtual venues, restricted user movement, generic rather than individualized configuration, and an expensive rehearsal space. This paper presents an Ambisonics method that addresses these limitations. The method simulates the acoustics of a chosen performance venue in real-time, by simulating oral-binaural room impulse responses (OBRIRs). This method allows changing the virtual venue and user-related data, and provides three-degrees-of-freedom (3DoF) for user head movement. The method is validated using quantitative and qualitative methods, and challenges of future real-time implementation are discussed. Despite the challenges, the method is capable of facilitating virtual rehearsal in real-time while providing for greater user flexibility.

This paper derives a superfast Toeplitz matrix inversion method using FFT techniques. The approach casts the non-periodic Toeplitz matrix solution into a periodic framework and uses the Woodbury matrix inverse identity to express the non-periodic solution as a correction to the time-windowed periodic solution. It is shown that the solution can be implemented using FFT techniques that avoid the calculation of a large matrix inverse. The method is implemented for both single- and multi-channel cases and typical computation times and accuracy are reported for the equalization of simulated and measured room impulse responses.

Impulse response shaping (IRS) is a prefiltering technique for modifying the characteristics of a linear channel to achieve desirable characteristics. Acoustic impulse response shaping is used to apply partial equalization to reduce the effects of reverberation on audio signals propagating inside a room and is thus used for listening room compensation. To be capable of responding rapidly to time varying room acoustics, a shaping algorithm must be computationally fast; this is difficult to achieve since the filters to be designed may have thousands of coefficients. This paper derives variants of the Dual Augmented Lagrangian Method, presents their application to the design of the shaping filters, and validates their efficacy through experimental studies. In addition, the formulations have been modified to design shaping filters that are robust to change in microphone positions.

Spatial audio may be performed in rooms using active-compensated sound field reproduction (AC-SFR), to cancel the reverberant reflections. Reproduction of a sound field may be performed using precalculated loudspeaker filters. However the robustness can be poor due to the temperature-induced perturbation of the room impulse responses (RIR). To compensate for reverberation, filters with significant non-causal components are designed. It is known that the requirements on the loudspeaker filters for room equalization may be reduced by applying the method of impulse response shaping (IRS). In this paper, we show that actively compensating paths with a long propagation distance are more heavily influenced by temperature than short paths. We hence impose a shaping penalty on the non-causal components of designed loudspeaker filters, to reduce the error caused by a change in temperature. We obtain AC-SFR designs that are more robust to changes in temperature.

Acoustic impulse response shaping is a prefiltering technique to reduce the perceptible effects of room reverberation. The room impulse responses to be shaped must first be measured, and these measurements can contain errors. Furthermore, room responses vary with changes in temperature and humidity and also with changes in measurement position. This letter presents a method for enhancing the robustness of the shaping procedure so that the effects of these errors and changes are minimized. The method uses a stochastic model of the channel variations to explicitly limit the probability of large deviations from the desired performance. The method is evaluated on realistic channel perturbations and the resulting shaped responses are shown to comply with the robustness specification.

In this paper, a pair of analytic expressions describing the correlation functions due to spherically symmetric radial Gaussian and uniform volume near-field source position distributions are presented. An approximate spatial correlation function solution for a radial Gaussian source location distribution is derived and compared with the existing numerical methods. An exact solution for a uniform volume source location distribution is also derived and compared with existing numerical methods. The approximate radial Gaussian solution produces a result consistent with numerical methods for compact source location distributions. The uniform volume solution matches the expected behavior. Finally, the spatial correlation functions were used to design spatially robust beamformers for compact microphone arrays. Both of the spatial correlation function solutions lead to improved spatial robustness for the applications of signal enhancement and suppression.

Cochlea is a part of the inner ear and it has complex anatomy and function. The proper functioning of the cochlea includes the generation of a traveling wave along the basilar membrane, which leads to depolarization of hair cells in the Organ of Corti and subsequent auditory nerve excitation (mechanoelectrical transduction). To represent the behavior of the organ of Corti, electro-mechanical cochlea model needs to be developed. This paper presents a simplified electro-mechanical state space model of the cochlea.

In this paper, a method of adaptive noise suppression combining spatially robust fixed beamforming and the TRINICON blind source separation algorithm is presented. A multichannel sensor array is first processed using complementary fixed beamformers into maximum and minimum SINR channels. The channels form the inputs to a single 2×2 second-order statistics TRINICON-BSS system which adaptively compensates for imperfections of the fixed beamformer design relative to the acoustic scenario. It is demonstrated that integrating the TRINICON-BSS algorithm leads to improved SINR performance over the initial imperfect beamformer design, and achieves a performance comparable to a perfect MVDR beamformer.

Impulse response shaping is a technique for partly equalizing impulse responses. In acoustics, it can be used for the reproduction of audio signals mitigated by distortions in a room. The most significant phenomenon among the distortions is reverberation, a straightforward characterization of which is the room impulse response. Room responses can be characterized but could contain measurement errors or noise. In addition, room responses vary with changes in atmospheric conditions such as temperature and humidity and also due to change in positions inside a room. The design of a shaping filter robust to at least some of these variations is likely to be very useful, which is considered in this work. The method uses a computationally efficient approach based on Basis Pursuit DeNoising (BPDN).

This paper presents spatially robust far-field microphone beamformers and nullformers derived using the von Mises and von Mises-Fisher distributions to model the expected direction of arrival. Simple analytic expressions are presented for 2D and 3D far-field correlation functions and used to design spatially robust beamformers and nullformers. It is demonstrated that the spatially robust beamformers show a modest improvement in tolerating uncertainty in the target direction of arrival without incurring a significant penalty in terms of SINR performance compared with the MVDR beamformer. In addition, the spatially robust formulation shows significantly improved numerical robustness, indicating improved ability in tolerating intrinsic array errors.

We investigate the prediction of wideband MIMO spatial channels. We propose a two-stage long range parametric prediction scheme that exploits the temporal, spatial and frequency correlations in a realistic cluster based fading channel. The proposed scheme utilizes the frequency correlation in an ESPRIT-like approach to estimate the cluster delays and scattering coefficients. The spatial and temporal correlations are then used to jointly estimate the angles of arrival, angles of departure and Doppler shifts via a 3D ESPRIT algorithm. Simulation results using the standardized 3GPP/WINNER II spatial channel model show that the proposed algorithm offers improved prediction performance over previous methods and can achieve longer prediction range.

A novel prediction scheme for polarized narrowband MIMO channels is proposed in this paper. The prediction scheme is based on estimation of the parameters of a double directional polarized propagation model. The proposed algorithm transforms the channel impulse response matrix in such a manner that a multidimensional extension of the ESPRIT algorithm can be utilized to jointly estimate the angles of arrival, angles of departure, Doppler shifts and complex polarimetric weights of the dominant multipath components. Simulation results show that the proposed algorithm outperforms repeated application of one dimensional ESPRIT based approach with similar computational complexity.

In this letter, we derive simple expressions for the lower bound on the prediction error variance for narrowband MIMO channel with uniform linear array at both ends of the link. The derived bounds show the relationship between the achievable prediction performance and prediction algorithm design parameters, thereby providing useful insights into the development of fading channel prediction algorithms.

We consider a cognitive radio (CR) relay network consisting of a cognitive source, a cognitive destination and a number of cognitive relay nodes that share spectrum with a primary transmitter and receiver. Due to poor channel conditions, the cognitive source is unable to communicate directly with the cognitive destination and hence employs the cognitive relays for assistance. We assume that perfect channel state information (CSI) for all links is not available to the CR. Under the assumption of partial and imperfect CSI at the CR system, we propose new robust CR cooperative relay beamformers where either the total relay transmit power or the cognitive destination signal-to-interference-and-noise ratio (SINR) is optimized subject to a constraint on the primary receiver outage probability. We formulate the robust total relay power minimisation and the cognitive destination SINR maximisation optimisation problems as a convex second order cone program and a semidefinite program, respectively. Cumulative distribution functions of primary receiver and cognitive destination receiver SINR for Rayleigh fading channels are presented.

The reweighing process of particle filtering can result in very large variance of particle weights; a problem known as degeneracy. The usual solution to this is an intermediary resampling step in which particles with lower weights are replaced by copies of those with large weights. This resampling inevitably results in loss of the information contained in those particles of low weights. Most of the existing stochastic and deterministic resampling schemes cause further loss of information because all the resampled particles have equal weights. These current techniques are what we would call, “hard resamplers,” and they impede the accumulation of information over many successive observations, which affects the detection of very covert targets. This paper presents two variants of a soft and deterministic resampling procedure that retains information contained in the weights over multiple observations. We demonstrate their effectiveness using a track-before-detect application.

We consider a cognitive radio (CR) relay network consisting of a cognitive source, a cognitive destination and a number of cognitive relay nodes that share spectrum with a primary transmitter and receiver. Due to poor channel conditions, the cognitive source is unable to communicate directly with the cognitive destination and hence employs the cognitive relays for assistance. We assume that the CR has a very loose cooperation with the primary network and therefore, only partial channel state information is available. Under these assumptions, we propose a new statistically robust CR cooperative relay beamformer where either the total relay transmit power or the cognitive destination signal-to-interference-and-noise ratio (SINR) is optimised subject to primary receiver outage probability constraint. We formulate the robust total relay power and the cognitive destination SINR optimisation problems as a convex second order cone program and a convex feasibility problem, respectively, that provide near optimum results. We also present efficient iterative algorithms that provide the optimum results.

In this paper, we propose an ESPRIT-based parametric prediction scheme for narrowband MIMO systems that fully exploits both temporal and spatial correlations in realistic MIMO channels. The proposed predictor uses a vector transmit spatial signature model and two-dimensional ESPRIT for the estimation of the channel parameters. The proposed scheme outperforms existing algorithms and is well suited to both two dimensional azimuth only and three dimensional MIMO spatial channel models.

An impulse response shortening approach is used to perform acoustic crosstalk cancellation. Crosstalk canceler filters are traditionally designed using least squares, with an approach that equalizes all room reverberation. However, depending upon end application, some reverberation may be permissible in the delivered signals. This idea is used to create more efficient crosstalk cancellation filters. The filter design is formulated as a minimax problem solvable with linear programming methods. Penalty functions on crosstalk levels and detrimental reverberation are introduced, which allow control of the reverberant tails and crosstalk levels. Shorter crosstalk cancellation filters are designed, by leaving in early echoes and/or allowing a slower decay of the late reverberant tail.

We consider a cognitive radio system with N secondary user (SU) pairs and a pair of primary users (PU). The SU power allocation problem is formulated as a rate maximisation problem under PU and SU quality of service and SU peak power constraints. We show our problem formulation is a geometric program and can be solved with convex optimisation techniques. We examine the effect of PU transmissions in our formulations. Solutions for both low and high signal-to-interference-and-noise ratio (SINR) scenarios are provided. We show that including the PU rate in the optimisation problem leads to increased PU performance while not significantly degrading SU rate. Achievable rate cumulative distribution functions for various Rayleigh fading channels are produced.

Cognitive Radio (CR) is a key technology for improving the utilization level of the overall radio spectrum in wireless communications. It is able to sense and change its transmission and reception parameters adaptively according to spectrum availability in different spectrum channels. The Cognition Cycle (CC) is a state machine that is embodied in each CR host that defines the mechanisms related to achieving context awareness and intelligence including observation, learning, and action selection. The CC is the key element in the design of various applications in CR networks such as Dynamic Channel Selection (DCS), scheduling and congestion control. In this paper, Reinforcement Learning (RL) is employed to implement the CC in mobile CR networks. Previous works consider static networks with homogeneous channels. This paper analyzes the performance of RL as an approach to achieve context awareness and intelligence in regard to DCS in mobile CR networks with heterogeneous channels. Our contribution in this paper is to show whether RL is an appropriate tool to implement the CC. The results presented in this paper show that RL is a promising approach.

A recent approach to surround sound is to perform exact control of the sound field over a region of space. Here, the driving signals for an array of loudspeakers are chosen to create a desired sound field over an extended area. An interesting subtopic is multi-zone surround sound, where two or more listeners can experience totally independent sound fields. However, multi-zone surround sound is a challenge because implementation can be very non-robust. We formulate multi-zone sound reproduction as a convex optimization problem, where the sound energy leakage into other listener zones is limited to fixed levels, and a constraint is placed on the loudspeaker weights to improve the robustness. An interior point algorithm is de vised for computing the loudspeaker weights, and its performance is compared with least squares approaches of multi-zone reproduction in typical two-zone cases.

Proyiding that licensed or Primary Users (PUs) are oblivious to the presence of unlicensed or Secondary Users (SUs), Cognitive Radio (CR) enables the SUs to use underutilized licensed spectrum (or white spaces) opportunistically and temporarily conditional on the interference to the PUs being below an acceptable level. Context awareness and intelligence enable the SU to sense for and use the underutilized licensed spectrum in an efficient manner. This paper investigates various learning mechanisms for achieving context awareness and intelligence with respect to Dynamic Channel Selection (DCS) in CR networks. The learning mechanisms are Adaptation (Adapt), Window (Win), Adaptation-Window (AdaptWin), and Reinforcement Learning (RL). The DCS scheme helps SU base station to select channel adaptively for data transmission to its SU host in static and mobile centralized CR networks. The purpose is to enhance quality of service, particularly throughput and delay (in terms of number of channel switches), in the presence of channel heterogeneity. Our contribution is to investigate simple and yet pragmatic learning mechanisms for CR networks. Simulation results reveal that RL, AdaptWin and Win achieve approximately similar and the best possible network performance, followed by Adapt, and finally Random, which does not apply learning and serves as baseline.

The reproduction of a sound field over a spatial region can require large output powers from the loudspeakers. In such a case, the sound field reproduction can be non-robust or cause loudspeaker failure. One method of limiting loudspeaker power is to regularize the solution. We instead assume each loudspeaker is capable of delivering a certain peak power, and find a solution minimizing the error between the created and desired field subject to this power constraint. We apply a power constraint at a single frequency and also constrain total output power summed across many frequencies. Rapidly converging interior point methods are derived and shown to possess similar robustness to the regularized solution.

Cognitive Radio (CR) is a next-generation wireless communication system that exploits underutilized licensed spectrum to optimize the utilization of the overall radio spectrum. A Distributed Cognitive Radio Network (DCRN) is a distributed wireless network established by a number of CR hosts in the absence of fixed network infrastructure. Context-awareness and intelligence are key characteristics of CR networks that enable the CR hosts to be aware of their operating environment in order to make an efficient and optimal joint action. Applying our extended Payoff Propagation (PP) mechanism in DCRN helps the CR hosts to achieve an efficient and optimal joint action in a cooperative and distributed manner through learning. The PP is suitable to be applied in most schemes in DCRN that requires context-awareness and intelligence such as Dynamic Channel Selection (DCS), scheduling, and congestion control. We investigate the performance of the PP in respect to DCS, and show that it is able to converge to an efficient and optimal joint action in a distributed manner including a DCRN with cyclic topology; furthermore we show that fast convergence is possible.

Cognitive Radio (CR) is a next-generation wireless communication system that enables unlicensed users to exploit underutilized licensed spectrum to optimize the utilization of the overall radio spectrum. A Distributed Cognitive Radio Network (DCRN) is a distributed wireless network established by a number of unlicensed users in the absence of fixed network infrastructure such as a base station. Context awareness and intelligence are the capabilities that enable each unlicensed user to observe and carry out its own action as part of the joint action on its operating environment for network-wide performance enhancement. These capabilities can be applied in various application schemes in CR networks such as Dynamic Channel Selection (DCS), congestion control, and scheduling. In this paper, we apply Reinforcement Learning (RL), including single-agent and multi-agent approaches, to achieve context awareness and intelligence. Firstly, we show that the RL approach achieves a joint action that provides better network-wide performance in respect to DCS in DCRNs. The multi-agent approach is shown to provide higher levels of stability compared to the single-agent approach. Secondly, we show that RL achieves high level of fairness. Thirdly, we show the effects of network density and various essential parameters in RL on the network-wide performance.

Traditional static spectrum allocation policies have been to grant each wireless service exclusive usage of certain frequency bands, leaving several spectrum bands unlicensed for industrial, scientific and medical purposes. The rapid proliferation of low-cost wireless applications in unlicensed spectrum bands has resulted in spectrum scarcity among those bands. Since most applications in Wireless Sensor Networks (WSNs) utilize the unlicensed spectrum, network-wide performance of WSNs will inevitably degrade as their popularity increases. Sharing of under-utilized licensed spectrum among unlicensed devices is a promising solution to the spectrum scarcity issue. Cognitive Radio (CR) is a new paradigm in wireless communication that allows sensor nodes as the unlicensed users or Secondary Users (SUs) to detect and use the under-utilized licensed spectrum temporarily. Given that the licensed or Primary Users (PUs) are oblivious to the presence of SUs, the SUs access the licensed spectrum opportunistically without interfering the PUs, while improving their own performance. In this paper, we propose an approach to build Cognitive Radio-based Wireless Sensor Networks (CR-WSNs). We believe that CR-WSN is the next-generation WSN. Realizing that both WSNs and CR present unique challenges to the design of CR-WSNs, we provide an overview and conceptual design of WSNs from the perspective of CR. The open issues are discussed to motivate new research interests in this field. We also present our method to achieving context-awareness and intelligence, which are the key components in CR networks, to address an open issue in CR-WSN.

Cognitive radio (CR) is a novel and promising paradigm for next-generation wireless communication. It is able to sense and change its transmission and reception parameters adaptively according to spectrum availability at different channels. The cognition cycle (CC) is a state machine that is embodied in each CR that defines the mechanisms related to achieving context-awareness and intelligence including observation, orientation, learning, planning, decision making, and action selection. The CC is the key element in the design of various schemes in CR networks such as dynamic channel selection (DCS), scheduling and congestion control. Hence, a good implementation of the CC is of paramount importance. In this paper, reinforcement learning (RL) is employed to implement the CC. The main focus is to analyze the performance of RL as an approach to achieving context-awareness and intelligence in regard to DCS. The contributions of this paper are twofold. Firstly, we seek to justify whether RL is an appropriate tool to implement the CC. Secondly, we seek to understand the effects of changes on RL parameters on network performance. In addition, we propose solutions for the problems associated with the application of RL in DCS. The results presented in this paper show that RL is a promising approach.

Previous work has indicated that a limitation on the low frequency performance of a circular microphone array for holographic sound field recording is phase mismatch between the microphones in the array. At low frequencies these variations become more significant than at mid-range and high frequencies because the high order phase mode responses are lower in amplitude. This paper investigates the possibility of performing a "self" calibration of a microphone array. The basis of the calibration is to estimate the location of one or more sources using mid-range frequencies and to use this source location information to perform correction to the array at low frequencies. This of course implies that the calibration is performed in a relatively anechoic environment, since multipath effects at widely differing frequencies are uncorrelated. Initial results confirm that a significant improvement in the array response is possible using this approach.

A method is derived for passively locating wide-band targets (typically acoustic targets) which may be moving at speeds sufficient to produce significant Doppler shift. The method involves a relatively straightforward adaptation of standard beam forming techniques. It is shown that conventional beam forming techniques have less discrimination in the direction of motion of the sources, whereas the proposed technique exhibits no such degradation. The derivative and Hessian of the likelihood function are derived. These may be used for locating the maximum likelihood solution or for deriving a Gaussian approximation to the likelihood function for particle filtering applications. The expressions are applicable for subsonic and supersonic sources. The computation required for implementing a system based on the model is at present prohibitive

A method is derived for passively locating wide-band targets (typically acoustic targets) which may be moving at speeds sufficient to produce significant Doppler shift. The method involves a generalisation of standard beam forming techniques. It is shown that conventional beam forming techniques have less discrimination in the direction of motion of the sources, whereas the proposed technique exhibits no such degradation. The derivative and Hessian of the likelihood function can be used for locating the maximum likelihood solution or for deriving a Gaussian approximation to the likelihood function for particle filtering applications. The expressions are applicable for subsonic and supersonic sources

A bound is derived for the extent to which narrowband knowledge of a multipath field inside a spatially limited region can be extrapolated beyond that region. The bound is independent of any model which may be applied to the multipath field, relying on the field being a solution to the homogeneous Helmholtz wave equation. The bound has implications for the extent and reliability of long-range channel prediction, particularly when the multipath is diffuse.

The well known results of the spatial correlation function for 2-dimensional and 3-dimensional diffuse fields of narrowband signals are generalised to the case of general distributions of scatterers. A method is presented which allows closed form expressions for the correlation function to be obtained for arbitrary scattering distribution functions. These closed form expressions are derived for a variety of commonly used scattering distribution functions.

The well-known results of the spatial correlation function for two-dimensional and three-dimensional diffuse fields of narrowband signals are generalized to the case of general distributions of scatterers. A method is presented that allows closed-form expressions for the correlation function to be obtained for arbitrary scattering distribution functions. These closed-form expressions are derived for a variety of commonly used scattering distribution functions.

It has long been known that the performance of a channel with intersymbol interference (ISI) not only depends on the signal to noise ratio, but on the actual impulse response coefficients as well. In this paper, a technique for approximating the performance of a communications channel in the presence of ISI is presented. It is assumed that the finite length impulse response of the channel is known, and that a maximum likelihood sequence estimation (MLSE) technique is used for equalisation. The technique consists of identifying for the modulation and impulse response duration, those few error events which are most likely to occur. This selection process is computationally intensive, but once it has been performed for a given modulation, the results can be used to enable very efficient approximation. Results are presented for some common modulation schemes.