As an essential part of the emerging Internet of Things, connected and autonomous vehicles (CAVs) have the potential to reshape future transportation systems and change the commute style in people’s everyday life. Among many vehicular on-board devices, radar system and vehicle-to-vehicle (V2V) communication system are two important pillars for the realization of CAVs. In this paper, the concept of coordinated CRC (C-CRC) is proposed to serve as a cognitive mediator for bridging vehicular radar and communication systems. By establishing a mutual-assistance relationship, C-CRC provides a new safety mechanism that allows one system to learn from and react to the risks that the other system has encountered. Simulation results have shown that the proposed method has desirable performance in face of motion perturbation and/or jamming attack under various scenarios.

Connected and autonomous vehicles (CAVs) and unmanned aerial vehicles (UAVs) are viewed as revolutionary technologies in the era of Internet of Things (IoT). However, both CAV and UAV can be exploited by potential adversaries and pose serious threats to the intelligent transportation system (ITS), such as damaging the vehicle-to-vehicle (V2V) communication. In this paper, we investigate the anti-jamming V2V communication in an integrated UAV-CAV network with hybrid attackers, which consist of a malicious CAV with intelligent jamming capability and a malicious UAV without. To solve this problem, we propose to use a unique research tool named cognitive dynamic system (CDS), and apply its function of cognitive risk control (CRC) to develop an effective countermeasure. In each perception-action cycle (PAC), the power control will always be performed by a legitimate transmitting vehicle; meanwhile, the process of channel selection only takes place if the risk level is evaluated as high after completing the power control. This kind of design that involves task-switching is inspired by the predictive-adaptation feature of the human brain. Simulation results have shown that the proposed method based on CRC is able to defend hybrid attackers effectively under various settings.

The idea of cognitive dynamic system (CDS) was first described in a Point of View article published ten years ago. As simple as it was then, it is remarkable how far that early work on the CDS has progressed to make a big difference to engineering. The objective of this article is to highlight the recently pioneered elements of the CDS for how to bring risk under control for physical systems in the presence of uncertainties.

This paper addresses a brand new way of thinking that distinguishes itself by integrating two entities, namely, dynamic spectrum management on the one hand and cognitive dynamic system (CDS) on the other. To be more precise, the CDS performs the role of a cognitive engine centered at the heart of wireless communications, wherein licensed bands rooted in network providers responsible for primary users and unlicensed bands rooted in other networks responsible for secondary users. Thus, these two worlds embody primary users and secondary users, which collectively function within a single complex wireless communication network empowered by the CDS for the first time ever.

The papers in this special issue focus on the topic of Big Data. This is a second in a series of papers on the topic of Big Data. This first part in this series (IEEE Proceedings, January 2016) consist of papers that are devoted to the theoretical aspects of big data. This second in a two-part series focuses on the practical applications of Big Data.

This special issue highlights a number of algorithmic approaches that are fundamental to data analysis, both in formulating and solving problems that relate to Big Data.

Briefly, main purpose of the paper is fourfold: a) Cognitive perception, which consists of two functional blocks: improved sparse-coding under the influence of perceptual attention for extracting relevant information from the observables and ignoring irrelevant information, followed by a Bayesian algorithm for state estimation. b) Entropic state of the perceptor, which provides feedback information to the controller. c) Cognitive control, which also consists of two functional blocks: executive learning algorithm computed by processing the entropic state, followed by predictive planning to set the stage for policy to act on the environment, thereby establishing the global perception-action cycle. d) Experimental results for exploiting the perceptual as well as executive attention in a co-operative manner, which is aimed at the first demonstration of risk control in the presence of a severe disturbance in the environment.

Sparse coding has established itself as a useful tool for the representation of natural data in the neuroscience as well as signal-processing literature. The aim of this letter, inspired by the human brain, is to improve on the performance of the sparse coding algorithm by trying to bridge the gap between neuroscience and engineering. To this end, we build on the localized perception-action cycle in cognitive neuroscience by categorizing it under the umbrella of perceptual attention, which lends itself to increase gradually the contrast between relevant information and irrelevant information. Stated in another way, irrelevant information is filtered away, while relevant information about the environment is enhanced from one cycle to the next. We may thus think in terms of the information filter, which, in a Bayesian context, was introduced in the literature by Fraser (1967). In a Bayesian context, the information filter provides a method for algorithmic implementation of perceptual attention. The information filter may therefore be viewed as the basis for improving the algorithmic performance of sparse coding. To support this performance improvement, the letter presents two computer experiments. The first experiment uses simulated (real-valued) data that are generated to purposely make the problem challenging. The second uses real-life radar data that are complex valued, hence the proposal to introduce Wirtinger calculus into derivation of the new algorithm.

The articles in this special issue focus on cognitive dynamic systems and applications for their use.

Cognitive dynamic systems provide a broadly defined platform, whereby engineering learns from cognitive neuroscience, and by the same token, cognitive neuroscience learns from engineering. The first part of the paper is of a tutorial nature, addressing recent advances in cognitive perception and cognitive control, which are the dual of each other. The study of cognitive perception, viewed from the perspective of Bayesian inference, starts with sparse coding, well known in neuroscience. However, sparse coding could become ill-posed, particularly when the signal-to-noise ratio is low. In such situations, stability is a necessary requirement, which can only be satisfied if there is sufficient information in the observables. To satisfy this requirement, the sparse-coding algorithm is augmented by the addition of information filtering (i.e., a special case of Bayesian filtering). Accordingly, the performance of sparse coding is improved under the influence of perceptual attention. This improvement enhances the cognitive perceptor to separate relevant information from irrelevant information. Next, moving into cognitive control, viewed from the perspective of Bellman's dynamic programming, two ideas are exploited: entropic state of the perceptor, and the definition of reward as an invertible function of two entropic states, namely, the current state and its immediate past value. The net result of building on these two ideas is a modified form of Bellman's dynamic programming, and, therefore, a new reinforcement learning algorithm, which not only outperforms traditional reinforcement learning algorithms, but also offers some highly desirable properties. Among them is a linear law of computational complexity, which is the best that it could be. The second part of the paper addresses two challenging problems: first, how to mediate between cognitive control and cognitive perception and, second, how to formulate a procedure for risk control. The first problem is resolved by making use of probabilistic reasoning, a branch of probability theory, which leads into the formulation of a probabilistic reasoning machine. With this mediation in place, the conditions for overall system stability are derived, thereby confirming the probabilistic reasoning machine as the overall system stabilizer. The second challenge is risk control, which is by far the most challenging of them all: In the presence of an unexpected disturbance in the environment, risk is brought under control by mimicking the predict and preadapt function, which is considered to be the overarching function in the prefrontal cortex of the brain. To be specific, motor control is expanded by the inclusion of a new preadaptive control mechanism, which involves two different sets of actions: One set is made up of possible actions identified by the policy in the motor control. The other set involves a window of experiences (i.e., optimal actions) gained in the past. In a novel way, by exploiting these two sets, we end up with a preadaptive control mechanism in the form of a closed-loop feedback structure, which brings with it control (executive) attention.

There are two worlds of wireless communications: the legacy wireless world and the cognitive wireless world. Spectrum holes are the medium through which the two worlds interact. Releasing subbands by primary users allows the cognitive radio users to perform their normal tasks and, therefore, to survive. In other words, the old world affects the new world through appearance and disappearance of the spectrum holes and there is a master-slave relationship between them. Hence, the two worlds of wireless communications are going on side by side. This makes a cognitive radio network a multiple-time-scale dynamic system: a large-scale time in which the activities of primary users change and a small-scale time in which the activities of secondary users change accordingly. Such systems are called double-layer dynamic systems. A model is built that can be used as a testing tool for policy forecast and incorporates time evolution as the life span (control horizon) of a given policy. Two types of time dependency are studied: time-dependent equilibria and time-dependent behaviour away from the predicted curve of equilibria. Theories of evolutionary variational inequalities and projected dynamic systems on Hilbert spaces are used to study these two types of time dependency, respectively.

Built on the theory of supply chain networks, this paper presents a model for the spectrum market, which includes three different tiers of decision makers; legacy owners, spectrum brokers, and secondary users (cognitive radios). Behavior of various decision-makers, the governing equilibrium conditions, and the transient behavior of the network are studied. Prices are determined endogenously in the model.

In this article on cognitive dynamic systems, the progress made and the way forward on this multidisciplinary integrative field has been reviewed. The following topics has been discussed: brief historical notes on human cognition; advances made on cognitive radar of the monostatic kind; emergence of cognitive control for the first time, building on the new concept of a two-state model and the significant progress made on cognitive radio on several fronts.

This paper is inspired by how cognitive control manifests itself in the human brain and does so in a remarkable way. It addresses the many facets involved in the control of directed information flow in a dynamic system, culminating in the notion of information gap, defined as the difference between relevant information (useful part of what is extracted from the incoming measurements) and sufficient information representing the information needed for achieving minimal risk. The notion of information gap leads naturally to how cognitive control can itself be defined. Then, another important idea is described, namely the two-state model, in which one is the system's state and the other is the entropic state that provides an essential metric for quantifying the information gap. The entropic state is computed in the perceptual part (i.e., perceptor) of the dynamic system and sent to the controller directly as feedback information. This feedback information provides the cognitive controller the information needed about the environment and the system to bring reinforcement leaning into play; reinforcement learning (RL), incorporating planning as an integral part, is at the very heart of cognitive control. The stage is now set for a computational experiment, involving cognitive radar wherein the cognitive controller is enabled to control the receiver via the environment. The experiment demonstrates how RL provides the mechanism for improved utilization of computational resources, and yet is able to deliver good performance through the use of planning. The paper finishes with concluding remarks.

In this paper, we describe a cognitive radar (CR) that mimics the visual brain. Although the visual brain and radar are different in that the visual brain does not transmit a probing signal to the environment while the active radar greatly relies on the probing signal it transmits to the environment, both of them are observers of the surrounding environment. As such, there is much that we can learn from the visual brain in building a new generation of CRs that outperform traditional radars. In this paper, we confine the discussion, in both analytic and experimental terms, to CR aimed at target tracking. From a theoretical perspective, using the posterior Cramér-Rao lower bound (PCRLB), it is shown that a cognitive tracking radar has the potential to improve tracking performance significantly. In particular, computer experiments are presented, which demonstrate that CR can indeed go beyond the theoretical limits of traditional active radars (TARs) as well as fore-active radars (FARs); the latter are radars equipped with feedback from the receiver to the transmitter. Moreover, computer experiments are presented to demonstrate another practical benefit resulting from the combined use of memory and executive attention in CR for a target-tracking application. Specifically, it is shown that with the provision of these two cognitive processes, the transition in switching from one transmit waveform to another goes forward in a smooth manner. Such a capability is beyond that of TAR or FAR.

This paper introduces the self-organizing dynamic spectrum management (SO-DSM) for ad hoc cognitive radio (CR) networks. In this scheme, CR units try to exploit the primary networks' unused bands by establishing links with neighbouring CRs using those bands. Inspired by human brain, the CRs extract and memorize the activity patterns of the primary network and other CRs and create temporal channel assignments on sub-bands with no recent primary user activities using a self-organizing maps (SOM) technique. The proposed scheme is decentralized and employs a simple learning rule with low complexity and minimal memory requirements. The simulation results show that SO-DSM significantly decreases the probability of collision with primary users and also probability of CR link interruption.

Much has been written on cognitive radio as a novel idea for improved utilization of the radio spectrum. Specifically, cognitive radio provides the means for this improved utilization by making it possible for secondary (cognitive radio) users to access subbands of the radio spectrum whenever and wherever the primary (legacy) users are not utilizing their subbands. It is, therefore, not surprising to find that papers written on cognitive radio have been growing exponentially in number over the past ten years. For a radio system to be cognitive in the full sense of the word, the five principles of cognition should feature in the design of the system. The five principles of cognition are as follows (Haykin, 2011): perception-action cycle; memory; attention; intelligence; and language.

Spectrum sensing is the very task upon which the entire operation of cognitive radio rests. For cognitive radio to fulfill the potential it offers to solve the spectrum underutilization problem and do so in a reliable and computationally feasible manner, we require a spectrum sensor that detects spectrum holes (i.e., underutilized subbands of the radio spectrum), provides high spectral-resolution capability, estimates the average power in each subband of the spectrum, and identifies the unknown directions of interfering signals. Cyclostationarity is another desirable property that could be used for signal detection and classification. The multitaper method (MTM) for nonparametric spectral estimation accomplishes these tasks accurately, effectively, robustly, and in a computationally feasible manner. The objectives of this paper are to present: 1) tutorial exposition of the MTM, which is expandable to perform space-time processing and time-frequency analysis; 2) cyclostationarity, viewed from the Loeve and Fourier perspectives; and 3) experimental results, using Advanced Television Systems Committee digital television and generic land mobile radio signals, followed by a discussion of the effects of Rayleigh fading.

A cognitive radio network is a multiuser system, in which different users compete for limited resources in an opportunistic manner, interacting with each other for access to the available resources. The fact that both users and spectrum holes (i.e., unused spectrum subbands) can come and go makes a cognitive radio network a highly dynamic and challenging wireless environment. Therefore, finding robust resource-allocation algorithms, which are capable of achieving reasonably good solutions fast enough in order to guarantee an acceptable level of performance under worst case interference conditions, is crucial in such environment. The focus of this paper is the transmit-power control in cognitive radio networks, considering a noncooperative framework. Moreover, tools from control theory are used to study both the equilibrium and transient behaviors of the network under dynamically varying conditions.

Abstract The currently in use spectrum management policies are responsible for the poor utilization of the electromagnetic radio spectrum. By performing dynamic spectrum management (DSM), cognitive radio (CR) has the potential to increase the radio spectrum efficiency significantly and has gained a lot of attention recently. In this paper, we present an overview of the DSM problem in CR. After describing the CR briefly, the DSM is explained. In order to increase the spectrum utilization efficiency, CR tries to share the spectrum with primary users. We discuss two methods for spectrum‐sharing, namely price‐based spectrum‐sharing and opportunistic spectrum‐sharing. After introducing necessary mathematical definitions, the formulation of the DSM problem is presented. We show that the DSM problem is equivalent to a well‐known graph‐coloring problem (GCP) called list‐coloring. Finding the exact solution for this problem is computationally intensive and various approximate algorithms have been proposed to obtain suboptimum solutions. Finally, we discuss two approaches for solving the DSM problem: centralized approach and decentralized approach. Decentralized approach, although has complicated design and may not achieve the global optimum solution, is more suitable for CR due to scalability and lower complexity. Copyright © 2009 John Wiley & Sons, Ltd.

The first half of the paper addresses the rationale for why we need to study cognitive dynamic systems, with particular reference to two wireless applications: cogitative radio for communication, and cognitive radar for remote sensing. The second half of the paper discusses the issues involved in dynamic spectrum management and transmit-power control, which are of particular importance to cognitive radio. The iterative water-filling algorithm, in a noncooperative radio environment is discussed, and its virtues and limitations are highlighted.

In this paper, a new version of the quadrature Kalman filter (QKF) is developed theoretically and tested experimentally. We first derive the new QKF for nonlinear systems with additive Gaussian noise by linearizing the process and measurement functions using statistical linear regression (SLR) through a set of Gauss-Hermite quadrature points that parameterize the Gaussian density. Moreover, we discuss how the new QKF can be extended and modified to take into account specific details of a given application. We then go on to extend the use of the new QKF to discrete-time, nonlinear systems with additive, possibly non-Gaussian noise. A bank of parallel QKFs, called the Gaussian sum-quadrature Kalman filter (GS-QKF) approximates the predicted and posterior densities as a finite number of weighted sums of Gaussian densities. The weights are obtained from the residuals of the QKFs. Three different Gaussian mixture reduction techniques are presented to alleviate the growing number of the Gaussian sum terms inherent to the GS-QKFs. Simulation results exhibit a significant improvement of the GS-QKFs over other nonlinear filtering approaches, namely, the basic bootstrap (particle) filters and Gaussian-sum extended Kalman filters, to solve nonlinear non- Gaussian filtering problems.

Multifunction radars (MFRs) are sophisticated sensors with complex dynamical modes that are widely used in surveillance and tracking. This paper demonstrates that stochastic context-free grammars (SCFGs) are adequate models for capturing the essential features of the MFR dynamics. Specifically, MFRs are modeled as systems that “speak” a language that is characterized by an SCFG. The paper shows that such a grammar is modulated by a Markov chain representing radar's policy of operation. The paper also demonstrates how some well-known statistical signal processing techniques can be applied to MFR signal processing using these stochstic syntactic models. We derive two statistical estimation approaches for MFR signal processing—a maximum likelihood sequence estimator to estimate radar's policies of operation, and a maximum likelihood parameter estimator to infer the radar parameter values. Two layers of signal processing are introduced in this paper. The first layer is concerned with the estimation of MFR's policies of operation. It involves signal processing in the CFG domain. The second layer is concerned with identification of tasks the radar is engaged in. It involves signal processing in the finite-state domain. Both of these signal processing techniques are important elements of a bigger radar signal processing problem that is often encountered in electronic warfare applications—the problem of the estimation of the level of threat that a radar poses to each individual target at any point in time.

Modeling wireless channels is essential to wireless communication systems. An autoregressive (AR) process of order one for wireless channel has long been assumed, but without a rigorous mathematical/physical basis. In this paper, we derive a first-order stochastic AR model for a flat stationary wireless channel, which comes from stochastic differential equation (SDE) theory concerning the nature of multipath fading channels. The resulting AR model describes more of the origin of multipath fading channels than previous AR models, and it can efficiently model and generate Rayleigh-distributed stationary fading channels. The Markovian property of the AR model is inherited through the SDE approach.

This paper investigates the improved decoder performance offered by incorporating dynamic linear modelling techniques when applied to particle filters for use in tracking the MIMO wireless channel. Conventional Bayesian-based receivers that perform channel tracking necessarily require a wireless channel model, typified by the use of a low order auto-regressive (AR) model. Normally, the model parameters are static in nature and are estimated a priori of any transmission; thus if the channel conditions change, a model mismatch occurs and system performance is degraded. Our method allows for time-varying channel statistics by modelling the channel fading rate as a Markov random walk. This new procedure allows the channel model to assume a time-varying behavior. As shown through simulations, the incorporation of dynamic modelling of time-dispersive channels not only offers superior performance, but at high SNR eliminates the error-rate floor commonly seen in systems using the static AR models

This paper highlights some fundamental issues involved in the study of large-scale dynamical systems. Two particular topics are discussed in some detail, one dealing with the management of active sensors via partially observable Markov decision processes, and the other dealing with the modeling, recognition and tracking of multi-function radars in an electronic warfare environment.

We present a new approach to radar pulse train analysis in electronic warfare. We consider an alternative to the classical time-of-arrival (TOA) histogram technique commonly used for extraction of complex pulse patterns. We derive a hidden Markov model for the radar word templates, and develop a modified version of the Viterbi algorithm to extract radar words from noisy and corrupted pulse sequences. We argue the advantages of this approach compared to the standard TOA histogram technique, and illustrate operation of the algorithm with computer simulation results.

Cognitive radio is viewed as a novel approach for improving the utilization of a precious natural resource: the radio electromagnetic spectrum. The cognitive radio, built on a software-defined radio, is defined as an intelligent wireless communication system that is aware of its environment and uses the methodology of understanding-by-building to learn from the environment and adapt to statistical variations in the input stimuli, with two primary objectives in mind: /spl middot/ highly reliable communication whenever and wherever needed; /spl middot/ efficient utilization of the radio spectrum. Following the discussion of interference temperature as a new metric for the quantification and management of interference, the paper addresses three fundamental cognitive tasks. 1) Radio-scene analysis. 2) Channel-state estimation and predictive modeling. 3) Transmit-power control and dynamic spectrum management. This work also discusses the emergent behavior of cognitive radio.

We propose a novel model-based hearing compensation strategy and gradient-free optimization procedure for a learning-based hearing aid design. Motivated by physiological data and normal and impaired auditory nerve models, a hearing compensation strategy is cast as a neural coding problem, and a Neurocompensator is designed to compensate for the hearing loss and enhance the speech. With the goal of learning the Neurocompensator parameters, we use a gradient-free optimization procedure, an improved version of the ALOPEX that we have developed, to learn the unknown parameters of the Neurocompensator. We present our methodology, learning procedure, and experimental results in detail; discussion is also given regarding the unsupervised learning and optimization methods.

This review presents an overview of a challenging problem in auditory perception, the cocktail party phenomenon, the delineation of which goes back to a classic paper by Cherry in 1953. In this review, we address the following issues: (1) human auditory scene analysis, which is a general process carried out by the auditory system of a human listener; (2) insight into auditory perception, which is derived from Marr's vision theory; (3) computational auditory scene analysis, which focuses on specific approaches aimed at solving the machine cocktail party problem; (4) active audition, the proposal for which is motivated by analogy with active vision, and (5) discussion of brain theory and independent component analysis, on the one hand, and correlative neural firing, on the other.

Recurrent neural networks (RNNs) have been successfully applied to communications channel equalization because of their modeling capability for nonlinear dynamic systems. Major problems of gradient-descent learning techniques commonly employed to train RNNs are slow convergence rates and long training sequences required for satisfactory performance. This paper presents decision-feedback equalizers using an RNN trained with Kalman filtering algorithms. The main features of the proposed recurrent neural equalizers, using the extended Kalman filter (EKF) and unscented Kalman filter (UKF), are fast convergence and good performance using relatively short training symbols. Experimental results for various time-varying channels are presented to evaluate the performance of the proposed approaches over a conventional recurrent neural equalizer.

We propose two novel Monte Carlo sampling-based Alopex (ALgorithm Of Pattern EXtraction) algorithms for training neural networks. The proposed algorithms naturally combine the sequential Monte Carlo estimation and Alopex-like procedure for gradient-free optimization, and the learning proceeds within the recursive Bayesian estimation framework. Experimental results on various problems show encouraging convergence results.

With the advent of cheap, low-power digital signal processors, it has been possible to develop hearing aids that utilize DSP technology. It is essential that the algorithms instantiated on these hearing aids be evaluated in a standard fashion with regards to both engineering and perceptual criteria. However, currently available hearing in noise tests do not allow for the range of signals or environments that are of interest to the designer of hearing aid algorithms. In addition, virtual audio environments have recently been suggested (Shinn-Cunningham, B.G., 20th Int. Conf. of the IEEE Engineering in Biology and Medicine Society, 1998) as a potential tool for the auditory science community. To address the needs of both engineering and clinical evaluation, we propose a flexible, software based virtual acoustic environment capable of realistically simulating a wide variety of scenarios.

This paper explores the use of particle filters, rooted in Bayesian estimation, as a device for tracking statistical variations in the channel matrix of a narrowband multiple-input, multiple-output (MIMO) wireless channel. The motivation is to permit the receiver to acquire channel state information through a semiblind strategy and thereby improve the receiver performance of the wireless communication system. To that end, the paper compares the particle filter as well as an improved version of the particle filter using gradient information, to the conventional Kalman filter and mixture Kalman filter with two metrics in mind: receiver performance curves and computational complexity. The comparisons, also including differential phase modulation, are carried out using real-life recorded MIMO wireless data.

Recurrent neural networks have been successfully applied to communications channel equalization because of their capability of modelling nonlinear dynamic systems. The major problems of gradient descent learning techniques, commonly employed to train recurrent neural networks, are slow convergence rates and long training sequences. This paper presents a decision feedback equalizer using a recurrent neural network trained with unscented Kalman filter (UKF). The main features of the proposed recurrent neural equalizer are fast convergence and good performance using relatively short training symbols. Experimental results for time-varying channels are presented to evaluate the performance of the proposed approaches over a conventional recurrent neural equalizer.

The implementation of current space-time codes is often performed under the assumption that the additive channel noise is white and Gaussian, and that the receiver has precise knowledge of the realization of the fading process. Here we study the application of particle filters to MIMO systems in order to reduce the uncertainty in the estimation of the channel fading coefficients. Using known orthogonal training sequences for channel estimation, an analysis on the estimated fading gains reveals that they are stochastic in nature with mean equal to the true channel gain, and variance proportional to the inverse of the transmit power. Furthermore, this estimation uncertainty is shown to incur a penalty in the signal-to-noise ratio, thus reducing overall system efficiency. In order to mitigate the effects of estimation error suffered by current MIMO systems, we use particle filters for channel tracking. Modeling the wireless fading channel as an AR process, the particle filter is shown to be superior to conventional estimation techniques by providing a significant decrease in the mean-squared error (MSE) of the channel estimate. Simulations illustrate the robust nature of this new scheme.

Simon Haykin Dept. of Electrical Engineering McMaster University [email protected] A major issue in evaluating speech enhancement and hearing compensation algorithms is to come up with a suitable metric that predicts intelligibility as judged by a human listener. Previous methods such as the widely used Speech Transmission Index (STI) fail to account for masking effects that arise from the highly nonlinear cochlear transfer function. We therefore propose a Neural Articulation speech intelligibility from the instantaneous neural spike rate over time, produced when a signal is processed by an auditory neural model. By using a well developed model of the auditory periphery and detection theory we show that human perceptual discrimination closely matches the modeled distortion in the instantaneous spike rates of the auditory nerve. In highly rippled frequency transfer conditions the NAI’s prediction error is 8% versus the STI’s prediction error of 10.8%.

Theoretical investigations of spatially correlated multitransmit and multireceive (MTMR) links show that not only independently and identically distributed links, but also spatially correlated links can offer linear capacity growth with increasing number of transmit and receive antennas. We explore the suitability of the turbo-BLAST architecture in correlated Rayleigh-fading MTMR environments. In particular, for an MTMR system with a large number of receive antennas, a near optimal performance can be achieved by the turbo-BLAST architecture in spatially and temporarily correlated Rayleigh-fading environments. The performance of turbo-BLAST, in terms of both bit-error rate and spectral efficiency, is analyzed empirically in indoors and correlated outdoor environments.

We consider the problem of designing a sparse FIR filter and show that it can be cast into a problem of determining a sparse solution of a linear system of equations. Previously proposed design algorithms for FIR filter utilize an intelligent search over all possible structures for sparse filter. We propose a new filter design method based on a simpler algorithm for finding a sparse solution of the linear system. Simulation experiments show significant improvements over classical nonsparse methods.

This review provides a comprehensive understanding of regularization theory from different perspectives, emphasizing smoothness and simplicity principles. Using the tools of operator theory and Fourier analysis, it is shown that the solution of the classical Tikhonov regularization problem can be derived from the regularized functional defined by a linear differential (integral) operator in the spatial (Fourier) domain. State-of-the-art research relevant to the regularization theory is reviewed, covering Occam's razor, minimum length description, Bayesian theory, pruning algorithms, informational (entropy) theory, statistical learning theory, and equivalent regularization. The universal principle of regularization in terms of Kolmogorov complexity is discussed. Finally, some prospective studies on regularization theory and beyond are suggested.

Nonlinear dynamics are basic to the characterization of many physical phenomena encountered in practice. Typically, we are given a time series of some observable(s) and the requirement is to uncover the underlying dynamics responsible for generating the time series. This problem becomes particularly challenging when the process and measurement equations of the dynamics are both nonlinear and noisy. Such a problem is exemplified by the case study of sea clutter which refers to radar backscatter from an ocean surface. After setting the stage for this case study, the paper presents tutorial reviews of: (1) the classical models of sea clutter based on the compound K distribution and (2) the application of chaos theory to sea clutter. Experimental results are presented that cast doubts on chaos as a possible nonlinear dynamical mechanism for the generation of sea clutter. Most importantly, experimental results show that on timescales smaller than a few seconds, sea clutter is very well described as a complex autoregressive process of order four or five. On larger timescales, gravity or swell waves cause this process to be modulated in both amplitude and frequency. It is shown that the amount of frequency modulation is correlated with the nonlinearity of the clutter signal. The dynamical model is an important step forward from the classical statistical approaches, but it is in its early stages of development.

TURBO-BLAST is a novel multi-transmit multi-receive (MTMR) antenna scheme for high-throughput wireless communications. It exploits a novel space-time coding scheme based on the independent block forward error correction (FEC) codes and space-time interleaving, and a near-optimal iterative decoder, for decoding a new generation of space-time codes. The proposed iterative decoder has two decoding stages: a soft interference cancelation detector and a set of soft-in soft-out decoders. We focus on designing a robust parallel interference cancelation scheme that jointly estimates the soft interference and the linear beamformer weights to minimize the mean-square error (MMSE) between the true and estimated signals. Using simulation results, we show that the proposed scheme outperforms the previously proposed soft interference cancelation receivers based on the maximum ratio combining (MRC) principle.

We propose a simplified Diagonal-BLAST (D-BLAST) architecture with parallel soft interference cancellation receiver based on the Turbo-BLAST (T-BLAST) architecture. In the T-BLAST architecture, the inter-substream coding is designed by a combination of random space-time interleaving and independent block encoding of each substream, using the same forward-error correction (FEC) code. We show that for the T-BLAST architecture, by using a systematic space-interleaving design that layers each substream diagonally across the antennas, a simplified diagonal inter-substream coding can be achieved without undue implementation complexity. The proposed diagonal inter-substream coding also facilitates the use of an iterative parallel interference cancellation receiver for decoding the simultaneously transmitted data, thereby achieving more capacity compared to the achievable capacity of traditional BLAST (Bell Labs Layered Space Time) architectures using sequential interference cancellation receivers. In this paper, we also present simulation results on fading channels, which confirm these findings.

The paper provides a new viewpoint on regularization theory from different perspectives. It is shown that the regularized solution can be derived from the Fourier transformation operator in the transformation domain and with equivalent form from the linear differential operator in the spatial domain. The state-of-the-art research in regularization is briefly reviewed with extended discussions on Occam's razor, minimum length description, Bayesian framework, pruning algorithms, statistical learning theory, and equivalent regularization.

This research monograph was developed in lectures to intermediate- level engineering and statistics students. It admirably surveys the field of feedforward networks considering both theoretical and practical aspects. The books attempts to respond to four fundamental questions concerning computational capabilities of feedforward networks:

We propose TURBO-BLAST, a novel multi-transmit, multi-receive antenna scheme, for high data rate wireless communications based on the Bell-Labs Layered Space Time (BLAST) architecture. In the TURBO-BLAST (T-BLAST) scheme, the incoming substreams use the same forward error correction (FEC) code, but they are interleaved differently using randomly generated inter-substream permuters. The transmitter structure leads to an iterative ("Turbo-like") receiver for decoding the simultaneously transmitted data. For the receiver, we consider the design and performance evaluation of an iterative and parallel soft interference-cancelation scheme followed by maximum-ratio combining (MRC) since it is simple and effective.

A new theory is developed for the feature spaces of hyperbolic tangent used as an activation kernel for non-linear support vector machines. The theory developed herein is based on the distinct features of hyperbolic geometry, which leads to an interesting geometrical interpretation of the higher-dimensional feature spaces of neural networks using hyperbolic tangent as the activation function. The new theory is used to explain the separability of hyperbolic tangent kernels where we show that the separability is possible only for a certain class of hyperbolic kernels. Simulation results are given supporting the separability theory.

Cloud base height is a continuous variable that falls within the range of zero to fourteen kilometers and is useful for understanding the Earth's radiation budget. Advances in LIDAR (laser radar) technology have provided accurate cloud base height measurements. However, new sensor development and deployment are costly processes. The paper is motivated by a desire to make LIDAR output of cloud base height information available at a network of ground based meteorological stations without actually installing LIDAR sensors. To accomplish this, fifty-seven sensors ranging from multispectral satellite information to standard atmospheric measurements such as temperature and humidity, are fused in what can only be termed as a very complex, non-linear environment. The result is an accurate prediction of cloud base height. Thus, a virtual sensor is created. This fusion is performed via neural network architectures. More specifically the choices of learning algorithms reflect the state-of-the-art in neural network design and include; as local methods, the regularized radial basis function (RBF) network and the support vector machine (SVM). Global methods include the node decoupled extended Kalman filter trained multi-layer perception (NDEKF-MLP), and as a benchmark, the venerable backpropagation algorithm. Overall, the support vector machine has shown itself to be the method of choice especially when complexity was considered, Excessive storage requirements occurred in the RBF case and the global methods required large committee machines to overcome the effects of local minima.

The support vector machine (SVM) constitutes one of the most powerful methods for constructing a mathematical model on the basis of a given number of training examples. SVM training requires that we solve a quadratic optimization problem; this step is usually performed by means of existing software packages. Such a black-box approach may be undesirable. In this paper we introduce a simple iterative algorithm for SVM training which compares well with some typical software packages, can be simply implemented, and has minimal memory requirements. It addresses the problem of regression estimation and utilizes ideas similar to those proposed by J. Platt (1998) for training binary SVM.

Most support vector (SV) methods proposed in the recent literature can be viewed in a unified framework with great flexibility in terms of the choice of the kernel functions and their constraints. We show that all these problems can be solved within a unique approach if we are equipped with a robust method for finding a sparse solution of a linear system. Moreover, for such a purpose, we propose an iterative algorithm that can be simply implemented. Finally, we compare the classical SV approach with other, recently proposed, cross-correlation based, alternative methods. The simplicity of their implementation and the possibility of exactly calculating their computational complexity constitute important advantages in a real-time signal processing scenario.

One of the fundamental issues in the operation of a mobile communication system is the assignment of channels to cells and to calls. Since the number of channels allocated to a mobile communication system is limited, efficient utilization of these communication channels by using efficient channel assignment strategies is not only desirable but also imperative. This paper presents a novel approach to solving the dynamic channel assignment (DCA) problem by using a form of realtime reinforcement learning known as Q-learning in conjunction with neural network representation. Instead of relying on a known teacher, the system is designed to learn an optimal channel assignment policy by directly interacting with the mobile communication environment. The performance of the Q-learning-based DCA was examined by extensive simulation studies on a 49-cell mobile communication system under various conditions. Comparative studies with the fixed channel assignment (FCA) scheme and one of the best dynamic channel assignment strategies, MAXAVAIL, have revealed that the proposed approach is able to perform better than the FCA in various situations and capable of achieving a performance similar to that achieved by the MAXIAVIAL, but with a significantly reduced computational complexity.

Concerns the improved detection of a nonstationary target signal in a nonstationary background. Ways to deal with the issue of nonstationarity are discussed, starting with Loeve's probabilistic theory of stationarity processes (1946, 1963). Three important tools emerge: the dynamic spectrum, the Wigner-Ville distribution as an instantaneous estimate of the dynamic spectrum and the Loeve spectrum. Procedures for the estimation of these spectra are described, and their applications are demonstrated using real-life radar data. Time, an essential dimension of learning, appears explicitly in the dynamic spectrum and Wigner-Ville distribution and implicitly in the Loeve spectrum. In each case, the 1D time series is transformed into a 2D image where the presence of nonstationarity is displayed in a more visible manner than in the original time series. This sets the stage for reformulating the signal detection problem as an adaptive pattern classification whereby we can exploit the learning property of neural nets. Hence, we describe a novel learning strategy for distinguishing between the different classes of received signals, such as 1) there is no target signal present in the received signal; 2) the target signal is weak; and 3) the target signal is strong. We present a case study based on real-life radar data. The case study demonstrates that the adaptive approach described is superior to the classical approach.

Oscillatory states in the electroencephalogram (EEG) reflect the rhythmic synchronous activity in large networks of neurons. Time-frequency (TF) methods, which quantify the spectral content of the EEG as a function of time, are well suited as tools for the study of spontaneous and induced changes in oscillatory states. The use of these methods provides insights into the temporal dynamics of EEG activity in both humans and experimental animals, and aids the study of the neuronal mechanisms that generate rhythmic EEG activity. Further the use of TF coherence analysts, which quantifies the consistency of phase relationships in multichannel EEG recordings, may contribute to the understanding of signal transmission between neuronal populations in different parts of the brain. We have used TF techniques to analyze the flow of activity patterns between two strongly connected brain structures: the entorhinal cortex and the hippocampus. Both of these structures are believed to be involved in information storage. By applying various frequencies of stimulation, we have found a peak in the spectral power in both sites at around 18 Hz, but the coherence between the EEG signals recorded from these sites was found to increase monotonically up to about 35 Hz. We have also found that long-term potentiation, a strong increase in the efficacy of excitatory synapses between these sites, either had no effect or decreased coherence.

A new method to construct a chaotic generator is presented. It is based on the fact that sea clutter is chaotic, and it can be reconstructed by the use of a supervised neural network. The constructed chaotic sequence generator is then applied to the direct-sequence coherent BPSK communication system as a pseudo-noise generator. Experimental results are presented. The advantages and disadvantages of the new pseudo-noise generator are discussed.

A dynamic network of regularized Gaussian radial basis functions (GaRBF) is described for the one-step prediction of nonlinear, nonstationary autoregressive (NLAR) processes governed by a smooth process map and a zero-mean, independent additive disturbance process of bounded variance. For N basis functions, both full-order and reduced-order updating algorithms are introduced, having computational complexities of O (N/sup 3/) and O (N/sup 2/), respectively, per time step. Simulations on a 10,000 point, 8-bit quantized 64 k bps rate speech signal show that the proposed dynamic algorithm has a prediction performance comparable and, in some cases, superior to that of AT&T's LMS-based speech predictor designed for the ITU-T G.721 standard on the 32 kbps ADPCM of speech. The results indicate that the proposed dynamic regularized GaRBF predictor provides a useful tradeoff between its minimal need for prior knowledge of the speech data characteristics and its consequently heavier computational burden.

This paper presents a tutor a overview of neural networks as signal processing tools for image compression. They are well suited to the problem of image compression due to their massively parallel and distributed architecture. Their characteristics are analogous to some of the features of our own visual system, which allow us to process visual information with much ease. For example, multilayer perceptions can be used as nonlinear predictors in differential pulse-code modulation (DPCM). Such predictors have been shown to increase the predictive gain relative to a linear predictor. Another active area of research is in the application of Hebbian learning to the extraction of principal components, which are the basis vectors for the optimal linear Karhunen-Loeve transform (KLT). These learning algorithms are iterative, have some computational advantages over standard eigendecomposition techniques, and can be made to adapt to changes in the input signal. Yet another model, the self-organizing feature map (SOFM), has been used with a great deal of success in the design of codebooks for vector quantization (VQ). The resulting codebooks are less sensitive to initial conditions than the standard LBG algorithm, and the topological ordering of the entries can be exploited to further increasing the coding efficiency and reducing the computational complexity.>

In this paper, we present a new method for the detection of signals in "noise", which is based on the premise that the "noise" is chaotic with at least one positive Lyapunov exponent. The method is naturally rooted in nonlinear dynamical systems and relies on neural networks for its implementation. We first present an introductory review of chaos. The subject matter selected for this part of the paper is written with emphasis on experimental studies of chaos using a time series. Specifically, we discuss the issues involved in the reconstruction of chaotic dynamics, attractor dimensions, and Lyapunov exponents. We describe procedures for the estimation of the correlation dimension and the largest Lyapunov exponent. The need for an adequate data length is stressed. In the second part of the paper we apply the chaos-based method to a difficult task: the radar detection of a small target in sea clutter.>

The optimal linear block transform for coding images is well known to be the Karhunen-Loeve transformation (KLT). However, the assumption of stationarity in the optimality condition is far from valid for images. Images are composed of regions whose local statistics may vary widely across an image. While the use of adaptation can result in improved performance, there has been little investigation into the optimality of the criterion upon which the adaptation is based. In this paper we propose a new transform coding method in which the adaptation is optimal. The system is modular, consisting of a number of modules corresponding to different classes of the input data. Each module consists of a linear transformation, whose bases are calculated during an initial training period. The appropriate class for a given input vector is determined by the subspace classifier. The performance of the resulting adaptive system is shown to be superior to that of the optimal nonadaptive linear transformation. This method can also be used as a segmentor. The segmentation it performs is independent of variations in illumination. In addition, the resulting class representations are analogous to the arrangement of the directionally sensitive columns in the visual cortex.

Proposes a novel approach to segmentation using a combination of Hebbian learning and competitive learning in a self-organizing manner. The network is modular, with each module corresponding to a different class of the input data. A module consists of a weight vector that is calculated during an initial training period. The appropriate class for a given input vector is determined by a maximum entropy classifier. The resulting network consistently extracts perceptually relevant features from image data. As well, the class representations are analogous to the arrangement of directionally sensitive columns in the visual cortex.>

In this paper we observe that a particular class of rational function (RF) approximations may be viewed as feedforward networks. Like the radial basis function (RBF) network, the training of the RF network may be performed using a linear adaptive filtering algorithm. We illustrate the application of the RF network by considering two nonlinear signal processing problems. The first problem concerns the one-step prediction of a time series consisting of a pair of complex sinusoid in the presence of colored non-gaussian noise. Simulated data were used for this problem. In the second problem, we use the RF network to build a nonlinear dynamic model of sea clutter (radar backscattering from a sea surface); here, real-life data were used for the study.

Parallel implementations of the extended square-root covariance filter (ESRCF) for tracking applications are developed. The decoupling technique and special properties used in the tracking Kalman filter (KF) are employed to reduce computational requirements and to increase parallelism. The application of the decoupling technique to the ESRCF results in the time and measurement updates of m decoupled (n/m)-dimensional matrices instead of one coupled n-dimensional matrix, where m denotes the tracking dimension and n denotes the number of state elements. The updates of m decoupled matrices are found to require approximately m fewer processing elements and clock cycles than the updates of one coupled matrix. The transformation of the Kalman gain which accounts for the decoupling is found to be straightforward to implement. The sparse nature of the measurement matrix and the sparse, band nature of the transition matrix are explored to simplify matrix multiplications.>

D.J. Thomson's method of multiple-windows (see Proc. IEEE, vol.70, p.1055-96, 1982) is applied to the problem of bearing (angle-of-arrival) estimation, at low grazing angles, in the combined presence of specular and diffuse multipath. The specular multipath component is coherently related to the direct signal component, both within a beamwidth of each other. They are modeled as line components in the wavenumber spectrum. The diffuse multipath component provides a colored noise background that is correlated to some extent with the line components. The estimator is tested with experimental data, collected by means of a 32-element sampled aperture antenna system operating at X-band over a lake surface. Additional improvement is observed when multiple frequencies and prior information is incorporated in the estimator.>

The authors briefly review the use of a random process and a dynamical system as mathematical models in the context of radar clutter. Experimental results are summarized for the correlation dimension of sea clutter, which provides an appropriate estimate for the embedding dimension of the process. The authors use the embedding dimension of sea clutter so estimated as a parameter in the design of a radial basis function (RBF) network for model reconstruction; here again experimental data are presented to support the theory.>

A general criterion is given to improve the accuracy of the predicted state x(k/k-1) in Kalman filter processing. The criterion is based on the orthogonal relation between the innovations process and past observations. Though this relation is basic to the operation of the Kalman filter, it is often not satisfied in the course of computation because of many target factors. The authors use this relation to construct a target function for minimizing the error. A nonlinear optimal algorithm, combining the standard Kalman filter and the target function equation, is formulated to process the target tracking problem. This algorithm is effective in decreasing the estimation error.>

The authors have developed an expansion they call the chirplet transform. It has been successfully applied to a wide variety of signal processing applications, including radar and image processing. There has been a recent debate as to the relative merits of an affine-in-time (wavelet) transform and the classical short-time Fourier transform (STFT) for the analysis of nonstationary phenomena. Chirplet filters embody both the wavelet and STFT as special cases by decoupling the filter bandwidths and center frequencies. Chirplets, by their embodiment of affine geometry in the time-frequency (TF) plane, may also include shears in time and frequency (chirps) and even time-bandwidth product variation (noise bursts) if desired. The most general chirplets may be derived from one or more basic ('mother') chirplets by the transformations or perspective geometry in the TF plane.>

By comparing the second-order statistics of the resolution cells under test with those obtained from their neighboring range cells, a detection scheme for detecting a target in a locally stationary two-dimensional random noise field is derived. The assumptions made are: (1) the noise field is uncorrelated in range but correlated in azimuth; (2) the correlation of the data in azimuth can be modeled by a Gaussian autoregressive process of relatively low order. The detection scheme exploits the full information of the second-order statistics in the data to make a detection decision. Its performance is superior to those schemes based on the amplitude or on Doppler shift information only. The proposed scheme is a constant false-alarm rate processor; its threshold can be evaluated by using a central chi-square probability distribution factor and its detection probability can be predicted by a non-central one.>

The results of an experimental study aimed at the classification of radar clutter encountered on ground-based coherent scanning radar systems used for air traffic control are presented. The clutter signals of interest are primarily those due to birds and to clouds and weather systems. A historical perspective on the radar clutter classification problem is given, and related issues are discussed. The important features of radar as a sensor in an air traffic control environment are described, and physical phenomena in radar clutter and targets, which provide the physical basis for the discrimination between the different radar clutter classes, are discussed. The feature of selection/extraction procedure, which is based on the multisegment Burg algorithm, is described. The experimental evaluation of a parametric Bayes classifier and a neural network classifier is reported.>

The important features of a coherent dual-polarized X-band radar designed to be of instrumentation quality for research use are described. The motivation for building the radar was to quantify the improvements attainable through the use of coherence and polarization in the detection of small floating targets in the presence of sea clutter. Results on the statistics of dual-polarized radar returns from the ocean surface (obtained with this radar at a site on the east coast of Canada) are presented. These results indicate that the K-distribution is very useful for describing the amplitude statistics of sea clutter (both the like- and cross-polarized channels) for low grazing angles. Analysis of the time-varying Doppler spectra of a small ice target and the neighboring sea clutter reveals differing Doppler characteristics, offering improved target detection.>

A classifier that incorporates both preprocessing and postprocessing procedures as well as a multilayer feedforward network (based on the back-propagation algorithm) in its design to distinguish between several major classes of radar returns including weather, birds, and aircraft is described. The classifier achieves an average classification accuracy of 89% on generalization for data collected during a single scan of the radar antenna. The procedures of feature selection for neural network training, the classifier design considerations, the learning algorithm development, the implementation, and the experimental results of the neural clutter classifier, which is simulated on a Warp systolic computer, are discussed. A comparative evaluation of the multilayer neural network with a traditional Bayes classifier is presented.

The authors describe a high-speed systolic-based computer system designed to process real-life radar data in near real time. At the heart of the system is a ten-cell Warp (systolic) machine. A digital recorder is used to input data to the system, and a color monitor displays the output of the processing algorithms. The system permits the processing of large volumes of radar data in real time, displays the results of advanced signal-processing algorithms, and has the effect of bringing a radar environment inside the laboratory for experimental studies or, equivalently, taking a sophisticated computer outside into an operational radar environment. The authors briefly describe research efforts built around this facility and highlight the elements of the powerful signal processing neural network software that are basic to their execution.>

A multilayer neural network classifier has been successfully implemented on a Warp systolic computer for distinguishing several major categories of radar returns: target (aircraft), weather, birds, and ground. The experimental results show that the neural network method is better than the traditional statistical method, which gives an average rate of 81.8% for classifying target, weather, and birds, in the same SNR range. The design of this neural classifier also suggests that the preprocessing and postprocessing procedures based on some prior information about the input data are very important for enhancing the classification performance

The radial basis function algorithm is used to classify impulse radar waveforms from asphalt-covered bridge decks. A brief description of the impulse radar scenario is given, and the radial basis function algorithm is summarized. The classification results obtained using the algorithm are presented. Excellent success was obtained, with classification accuracies up to 99%. Training the radial basis function classifier is faster and less complex than training a back-propagation-based network, since the simple least-mean-squares (LMS) algorithm can be used to obtain estimates of the optimum weight values

A theoretical evaluation of the performance limits of the innovations-based detection algorithm, which provides an adaptive algorithm for discriminating between two hypotheses parameterized as autoregressive (AR) models, is presented. Two particular cases are considered. One pertains to a situation in which the variance of the interference (modeled as an AR process) acting alone is known a priori. The other pertains to a situation in which no such knowledge is available. Detailed computer simulations are carried out to confirm the practical validity of the theory.>

A parallel implementation of the extended Kalman filter (KF) for target tracking systems is presented. The decoupling technique is utilized to reduce computational requirements and to increase parallelism. Also, the properties of matrices in the tracking KF are exploited to simplify the implementation complexity.>

The results of an exhaustive computer search for linear, nonredundant minimum missing-lag arrays are presented. The procedure provides arrays having the most densely packed autocorrelation estimates under the constraint that each estimate be at a distinct correlation lag.

The paper examines the use of a surface-based marine radar for imaging the surface features of sea ice in full ice cover. Two aspects of the problem are considered: 1) selection of radar parameters which maximize the effectiveness of the radar for ice imaging, and 2) evaluation of image processing (display) techniques that effectively convey to an operator the information contained in the radar returns. Both aspects of the study are supported by data collected at a radar site on Baffin Island, Canada. The paper also briefly describes the characteristics of first-year ice, multiyear ice, and icebergs, which are pertinent to the understanding of the radar ice-imaging problem. Based on the results presented on radar parameter evaluation, it is recommended that: 1) the radar resolution should be as high as possible; 2) the antenna should be installed as high as possible; 3) the radar should receive both like- and cross-polarized returns; 4) two radars, operating at widely separate frequencies, should be used. Based on the results of the display work, it is recommended that: 1) the display should use a scan technology that affords the required number of intensity levels; 2) median filtering should be used to reduce the effects of spike noise, if present; 3) normalization should be used to remove the range-dependence of radar returns; 4) selective edge enhancement be employed to maximize significant image detail; 5) to fully utilize the intensity levels available in the display, the histogram of the image should be modified, taking into account the visual response of the operator; 6) color can be used to accommodate the display of multi-parameter images, such as those arising from the use of like. and cross-polarization; 7) Features 3), 4), and 5) may be integrated into a modular image processing system.

A method of simultaneously resolving both range and doppler ambiguities in a pulse-doppler radar is presented. The improvements over previous techniques are that only two PRFs are required and that the same number of doppler filters in each PRF is permitted.

This paper presents the results of a performance evaluation that were made on two angle of target arrival estimation schemes that are suitable for use in a low-angle tracking radar environment. The two schemes were: (1) A maximum likelihood receiver, and (2) an adaptive cancellation scheme supplied with a calibration curve. The paper also describes a sampled aperture radar facility that has been developed at the Communications Research Centre (Ottawa) for multipath radar studies. The facility was used to obtain multipath data for evaluating the performance of these two estimators. The results show that, under the right conditions, both processors are capable of resolving an angle of target arrival equal to a small fraction of a beamwidth, using one snapshot of array data. The results also show that the adaptive cancellation scheme, consistently, has a slightly better performance than the maximum likelihood receiver.

In this paper, the Kalman filter theory is used to develop an algorithm for updating the tap-weight vector of an adaptive tapped-delay line filter that operates in a nonstationary environment. The tracking behaviour of the algorithm is discussed in detail. Computer simulation experiments show that this algorithm, unlike the exponentially weighted recursive least-squares (deterministic) algorithm, is always stable. Simulation results are included in the paper to illustrate this phenomenon.

This paper presents the results of a preliminary experimental examination of the response of an adaptive lattice-structure prediction-error filter to non-stationary inputs. The lattice-structure examined in this study uses the standard gradient method to continuously implement Burg's harmonic-mean algorithm. The results show a basic relationship between filter response, depth and rate of modulation, and filter adaptive constant and order.

A rapidly convergent approximation to the likelihood ratio for the generic problem of detecting a stationary discrete-time stochastic process in additive white Gaussian noise is formulated. The derivation is based on the innovations approach to detection theory.

The adaptive lattice prediction-error filter has been proposed as an alternative for a fixed moving-target indicator (MTI) filter presently used to detect a signal returned from a moving target in the presence of non-moving or slowly moving clutter as encountered in a radar environment. This paper examines the results from applying several forms of the lattice filter to actual data from an operational surveillance radar under a wide variety of environmental conditions. These results are compared with those obtained using the existing fixed filter for this radar (with the same input) and conclusions on their relative suitability are presented.

This paper describes an adaptive interference cancelling procedure based on the Kalman filtering theory. The procedure is shown to have a rapid convergence rate and provide a high signal-to-interference ratio (SIR) when dealing with sources of interference located in the sidelobe region of an array antenna. The procedure also deals successfully with interferences even when they are separated from the target of interest by a very small fraction of a standard beamwidth. In such a case, however, the convergence process tends to be somewhat slow, depending on the SIR and number of adapted weights. To constrain the processor from responding to a target signal, a set of orthogonal beams are used to provide the desired input data, and the main beam is excluded from the procedure for calculating the weights of the adaptive interference canceller.

A generalization of Burg's algorithm is described for the maximum-entropy (ME) field-mapping of a space-time series that is characterized by a non-Toeplitz cross-spectral (CS) matrix. Stability of the original Burg algorithm is retained, that is, the reflection coefficient is less then unity in magnitude at each recursion of the algorithm. Examples are included to illustrate applicability of the generalized algorithm to spatial mapping of correlated sources, as encountered in an under-water acoustic environment.

This paper presents a comparison of four different algorithms for calculating the reflection coefficients of a lattice-structure prediction-error filter. This filter structure is gaining increasing use in the fields of adaptive filtering and linear prediction analysis. Comparison is made of the harmonic-mean (also known as Burg's algorithm), the geometric-mean, the forward-and-backward, and the forward/backward-minimum algorithms. Also compared are two methods of implementing the algorithms as recursive, continuously adaptive calculations. Results using simulated radar returns as the test signal are presented.

The paper presents some experimental results on the maximum entropy method (MEM) used to estimate the wavenumber spectrum produced by plane waves which are incident upon a receiving array antenna, as encountered in a low-angle tracking radar. The results show that, unless the direct and reflected signals in such a system are spatially uncorrelated, the use of the MEM in this application yields erroneous results.

An adaptive digital filtering scheme is presented which uses a lattice structure for adaptive prediction and elimination of radar clutter, using Burg's algorithm for the computation of the lattice coefficients. This method computes a minimum-phase prediction error filter directly from the radar data, while not having the end-bias problems common to many filtering schemes. This permits quick adaptation to changing clutter conditions. An integration decay constant allows the filter to adapt to longer duration signals (clutter) while passing shorter duration signals (targets). The filter design is described and experimental results are discussed.

A 36-channel digital sonar processor has been designed and constructed based on computer simulation results. For CW pulse operation, 16 stored doppler replicas in ROM units are used for correlation with the received signal. For FM pulse operation, a hard limited replica stored in a ROM is used to correlate with this received signal. The advantage of processing the signals in this way is that the replicas can be adjusted to change the effective filtering characteristic. For example, increasing the length of the stored replica reduces the effective bandwidth. Also, replicas can be changed with range or provide different doppler selections for different beams. Experimental results using ideal and real signals are provided.

A brief review of the maximum entropy method for spectral analysis of complex signals is presented. It is shown that, for the spectral analysis of radar clutter, reliable short-term spectral estimates can be obtained with a small number of data points, while keeping a good resolution capability.

A new technique for achieving system synchronization to the switching satellite is described; this technique requires only three earth stations, designated as control stations, to transmit synchronization signals. The three signals are used to measure the three space delays from the control stations to the satellite. The control stations broadcast the space delays to all other stations in the system. Thus, any other earth station can obtain synchronization passively by calculating its own space delay.

A computer simulation study has been performed to evaluate the performance of a variety of digital modulation techniques in transmitting high-rate digital data over a satellite channel. Results are presented showing comparative performance of various techniques in the form of error-rate curves and signal-to-noise ratio (SNR) degradation curves.

This article is concerned with the video signal obtained when a coherent pulsed radar scans through moving scintillating scatrefers (clutter). In particular, a formula is obtained for the autocorrelation function of the clutter, assuming Gaussian distributions of doppler and scintillation for the scatterers. Results are included on a computer model developed to simulate clutter with predefined characteristics.

A coarse search synchronization technique is described for switched time division multiple access (TDMA) satellite systems, using a coded search signal. It is shown that this method achieves coarse search synchronization much faster, on the average, than the burst synchronization method described by Rapuano and Shimasaki. A description of an integrated synchronization system that combines the coded search signal used for coarse synchronization with the sync burst signal used for fine adjustment and tracking is included.