Recently, a shift has been observed towards the so-called edge machine learning, which allow multiple devices with local computational and storage resources to collaborate with the assistance of a centralized server. The well-known federated learning approach is able to utilize such architectures by allowing the exchange of only parameters with the server, while keeping the datasets private to each contributing device. In this work, we propose a communication-efficient, fully distributed, diffusion-based learning algorithm that does not require a parameter server and propose an adaptive combination rule for the cooperation of the devices. By adopting a classification task on the MNIST dataset, the efficacy of the proposed algorithm is demonstrated in non-IID dataset scenarios.

In this paper, a novel technique for the hyperspectral image deconvolution problem is developed. First, considering the highly ill-posed nature of the examined problem, it is imperative to incorporate proper priors (regularizers) to capture the strong spectral and spatial dependencies of the hyperspectral images. Then, in light of this, a novel optimization problem is proposed by employing a convolutional neural network to act as a regularizer, which is learnt to reflect the properties of the signals of interest. To solve the proposed optimization problem, we use the half quadratic splitting methodology, thus designing an efficient iterative solver (iteration map). Based on the Deep Equilibrium (DEQ) modeling, which aims to express the proposed iterative solver as an equilibrium (fixed-point) computation, a highly interpretable deep learning-based network is derived, which can be trained endto-end. Extensive numerical results using two publicly available datasets illustrate that the proposed method markedly outperforms other state-of-the-art approaches.

In this paper, we propose a novel cooperative-based system that facilitates each autonomous vehicle of the swarm to be fully aware of its 5 degrees of freedom (DOF) motion, i.e., 3D translation and 2D rotation, a very important task for autonomous navigation, known also as simultaneous localization and mapping (SLAM). The novelty is that the interconnected vehicles of the swarm share a common collective task: simultaneously estimating self and neighboring vehicles’ 5 DOF by perceiving, transmitting, associating and fusing heterogeneous data, e.g., visual, mechanical, satellite based, etc., relying on different sensor modalities and vehicular communication. The proposed sensor fusion framework is based on the Extended Kalman Filter algorithm, which is reformulated in order to capture cooperative 3D translation and 2D rotation estimation in an alternating fashion. Numerical results using the driving parameters of many cars from CARLA simulator, indicate very promising accuracy in terms of absolute trajectory error.

In this study, the problem of computing a sparse representation of multi-dimensional visual data is considered. In general, such data e.g., hyperspectral images, color images or video data consists of signals that exhibit strong local dependencies. A new computationally efficient sparse coding optimization problem is derived by employing regularization terms that are adapted to the properties of the signals of interest. Exploiting the merits of the learnable regularization techniques, a neural network is employed to act as structure prior and reveal the underlying signal dependencies. To solve the optimization problem Deep unrolling and Deep equilibrium based algorithms are developed, forming highly interpretable and concise deep-learning-based architectures, that process the input dataset in a block-by-block fashion. Extensive simulation results, in the context of hyperspectral image denoising, are provided, which demonstrate that the proposed algorithms outperform significantly other sparse coding approaches and exhibit superior performance against recent state-of-the-art deep-learning-based denoising models. In a wider perspective, our work provides a unique bridge between a classic approach, that is the sparse representation theory, and modern representation tools that are based on deep learning modeling.

Cooperative Intelligent Transportation Systems envision the integration of cooperative intelligence as a key operational part of autonomous driving. In this way, a fleet or swarm of Connected and Automated Vehicles collectively coordinates its driving actions in order to maximize its performance. To realize this ambition, vehicles need to be fully location-aware of their surrounding environment, through distributed AI intelligence. Motivated by this requirement, we develop in this paper a distributed cooperative awareness scheme which performs multi-modal fusion of heterogeneous sensor sources along with V2V communication information, using graph Laplacian matrix and Least-Mean-Squares algorithm. The intuition behind our approach is that neighboring vehicles are interested in estimating common positions of other vehicles. We build upon our previous work on global awareness though local information diffusion, and prove that the proposed distributed framework is able to address highly efficient the case of lacking any information about other networked vehicles. More specifically, our approach achieves high enough convergence speed as well as location accuracy. The evaluation study has been performed in CARLA autonomous driving simulator and verifies the proposed method’s benefits over other related solutions.

In this paper, we design distributed multi-modal localization approaches for Connected and Automated vehicles. We utilize information diffusion on graphs formed by moving vehicles, based on Adapt-then-Combine strategies coupled with the Least-Mean-Squares and the Conjugate Gradient algorithms. We treat the vehicular network as an undirected graph, where vehicles communicate with each other by means of Vehicle-to-Vehicle communication protocols. Connected vehicles perform cooperative fusion of different measurement modalities, including location and range measurements, in order to estimate both their positions and the positions of all other networked vehicles, by interacting only with their local neighborhood. The trajectories of vehicles were generated either by a well-known kinematic model, or by using the CARLA autonomous driving simulator. The proposed distributed and diffusion localization schemes significantly reduced the GPS error and do not only converged to the global solution, but they even outperformed it. Extensive simulation studies highlight the benefits of the various methods, which in turn outperform other state of the art approaches. The impact of the network connections and the network latency are also investigated.

Given two datasets that belong to different feature spaces and both correspond to the same underlying phenomenon, the scope of coupled dictionary learning is to compute two dictionaries, one for each dataset, so that each dataset is approximated using the respective dictionary but the same sparse coding matrix. In this work, the focus is on a particular, yet widespread, form of this problem in which the datasets correspond to slowly varying (piece-wise smooth) signals, and the measurements contain severe noise. A novel coupled dictionary learning technique is developed by including a suitable total-variation-based regularization term in the cost function. Furthermore, exploiting the smoothness of the datasets, new fast sparse coding algorithms are derived. The new techniques achieve effective modeling of the smooth signal and significantly alleviate the effects of noise. Finally, extensive simulation results for the problem of spectral super-resolution of hyperspectral images are provided, demonstrating the performance improvements offered by the derived techniques.

In recent years, several successful schemes have been proposed to solve the song identification problem. These techniques aim to construct a signal's audio-fingerprint by either employing conventional signal processing techniques or by computing its sparse representation in the time-frequency domain. This paper proposes a new audio-fingerprinting scheme which is able to construct a unique and concise representation of an audio signal by applying a dictionary, which is learnt here via the well-known K-SVD algorithm applied on a song database. The promising results which emerged while conducting the experiments suggested that, not only the proposed approach preformed rather well in its attempt to identify the signal content of several audio clips -even in cases this content had been distorted by noise - but also surpassed the recognition rate of a Shazam-based paradigm.

The Industrial Internet of Things (IIoT) is a key element of industry 4.0, bringing together modern sensor technology, fog and cloud computing platforms, and artificial intelligence to create smart, self-optimizing industrial equipment and facilities. Though, the scale and sensitivity degree of information continuously increases, giving rise to serious privacy concerns. The scope of this article is to provide efficient privacy preservation techniques, by tracking the correlation of multivariate streams recorded in a network of IIoT devices. The time-varying data covariance matrix is used to add noise that cannot be easily removed by filtering, generating obfuscated measurements and, thus, preventing unauthorized access to the original data. To improve communication efficiency between connected IoT devices, we exploit inherent properties of the correlation matrices, and track the essential correlations from a small subset of correlation values. Extensive simulation studies using constrained IIoT devices validate the robustness, efficiency, and effectiveness of our approach.

Distributed estimation of a parameter vector in a network of sensor nodes with ambiguous measurements is considered. The ambiguities are modelled by following a set-theoretic approach, that leads to each sensor employing a non-convex constraint set on the parameter vector. Consensus can be used to reach an estimate consistent with the measurements of all nodes, assuming that such an estimate exists, but unfortunately, such an approach leads to a non-convex problem. Using proper assumptions, the considered problem is decomposed into two sub-problems, where one is well studied in literature and the other is modelled as a non-cooperative game. An exact potential function is derived for this game, and an algorithm for its solution is given. Numerical results, consistent with the theoretical findings, demonstrate the efficacy of the proposed approach.

The Industrial IoT (IIoT) is a key element of Industry 4.0, bringing together modern sensor technology, fog - cloud computing platforms, and artificial intelligence (AI) to create smart, self-optimizing industrial equipment and facilities. Though, the scale and sensitivity degree of information continuously increases, giving rise to serious privacy concerns. In this work we address the problem of efficiently and effectively tracking the structure of multivariate streams recorded in a network of IIoT devices. The time varying correlation data values are used to add noise which maximally preserves privacy, in the sense that it is very hard to be removed. To improve communication efficiency between connected IoT devices, we exploit low rank properties of the correlation matrices, and track the essential correlations from a small subset of correlation values estimated by a subset of network nodes. Extensive simulation studies, validate the correctness, efficiency, and effectiveness of our approach in terms of computational complexity, transmission energy efficiency and privacy preservation.

Recently, there has been increasing interest for easy and reliable generation of 3D animated models facilitating several real-time applications. In most of these applications, the reconstruction of soft body animations is based on time-varying point clouds which are irregularly sampled and highly incomplete. To overcome these imperfections, we introduce a novel reconstruction technique, using graph-based matrix completion approaches. The presented method exploits spatio-temporal coherences by implicitly forcing the proximity of the adjacent 3D points in time and space. The proposed constraints are modeled by using the weighted Laplacian graphs and are constructed from the available points. Extensive evaluation studies, carried out using a collection of different highly-incomplete dynamic models, verify that the proposed technique achieves plausible reconstruction output despite the constraints posed by arbitrarily complex and motion scenarios.

In this paper, the parameter estimation problem based on diffusion least mean squares strategies is studied from a coalitional game theoretical perspective. The problem has been modeled as a non-transferable coalitional game and two scenarios have been considered, one where the value function includes only a suitable estimation accuracy criterion and another one in which the cost of coalition formation is taken into account as well. In the former scenario, we first analyze the non-emptiness of the core of the games corresponding to traditional diffusion strategies, and then, we extend the analysis to a recently proposed node-specific parameter estimation setting where the nodes have overlapped but different estimation interests. In the latter scenario, we employ a distributed coalition formation algorithm, based on merge-and-split steps, which converges to a stable coalition structure.

In this paper, we study the problem of resource allocation for a multiuser orthogonal frequency-division multiple access (OFDMA) downlink with eavesdropping. The considered setup consists of a base station, several users, and a single eavesdropper that intends to wiretap the transmitted message within each OFDMA subchannel. By taking into consideration the existence of the eavesdropper, the base station aims to assign subchannels and allocate the available power in order to optimize the max-min fairness criterion over the users' secrecy rate. The considered problem is a mixed integer nonlinear program. For a fixed subchannel assignment, the optimal power allocation is obtained by developing an algorithm of polynomial computational complexity. In the general case, the problem is investigated from two different perspectives due to its combinatorial nature. In the first, the number of users is equal or higher than the number of subchannels, whereas in the second, the number of users is less than the number of subchannels. In the first case, we provide the optimal solution in polynomial time by transforming the original problem into an assignment one for which there are polynomial time algorithms. In the second case, the secrecy rate formula is linearly approximated and the problem is transformed to a mixed integer linear program, which is solved by a branch-and-bound algorithm. Moreover, optimality is discussed for two particular cases where the available power tends to infinity and zero, respectively. Based on the resulting insights, three heuristic schemes of polynomial complexity are proposed, offering a better balance between performance and complexity. Simulation results demonstrate that each one of these schemes achieves its highest performance at a different power regime of the system.

In this paper, we theoretically investigate a new technique for simultaneous information and power transfer (SWIPT) in multiple-input multiple-output (MIMO) point-to-point with radio frequency energy harvesting capabilities. The proposed technique exploits the spatial decomposition of the MIMO channel and uses the eigenchannels either to convey information or to transfer energy. In order to generalize our study, we consider channel estimation error in the decomposition process and the interference between the eigenchannels. An optimization problem that minimizes the total transmitted power subject to maximum power per eigenchannel, information and energy constraints is formulated as a mixed-integer nonlinear program and solved to optimality using mixed-integer second-order cone programming. A near-optimal mixed-integer linear programming solution is also developed with robust computational performance. A polynomial complexity algorithm is further proposed for the optimal solution of the problem when no maximum power per eigenchannel constraints are imposed. In addition, a low polynomial complexity algorithm is developed for the power allocation problem with a given eigenchannel assignment, as well as a low-complexity heuristic for solving the eigenchannel assignment problem.

Eigenvalue-based spectrum sensing (EBSS) techniques may operate in a totally blind manner while they offer remarkably improved performance for specific types of signals compared with energy-based methods. In the literature so far, only batch and centralized cooperative EBSS techniques have been considered, which, however, suffer from limitations that render them impractical in several cases. Thus, the aim of this paper is to develop practical cooperative adaptive versions of typical EBSS techniques that could be applied in a completely decentralized manner. To this end, at first, novel adaptive EBSS techniques are developed for the maximum eigenvalue detector, the maximum-minimum eigenvalue detector, and the generalized likelihood ratio test scheme, respectively, for a single-user (no cooperation) case. Then, a novel distributed subspace tracking method is proposed, which enables the cooperating nodes to track the joint subspace of their received signals. Based on this method, cooperative decentralized versions of the adaptive EBSS techniques are subsequently developed that overcome the limitations of the existing batch centralized approaches. The performance of the proposed methods is verified via indicative simulations.

A distributed adaptive algorithm is proposed to solve a node-specific parameter estimation problem where nodes are interested in estimating parameters of local interest and parameters of global interest to the whole network. To address the different node-specific parameter estimation problems, this novel algorithm relies on a diffusion-based implementation of different Least Mean Squares (LMS) algorithms, each associated with the estimation of a specific set of local or global parameters. Although all the different LMS algorithms are coupled, the diffusion-based implementation of each LMS algorithm is exclusively undertaken by the nodes of the network interested in a specific set of local or global parameters. To illustrate the effectiveness of the proposed technique we provide simulation results in the context of cooperative spectrum sensing in cognitive radio networks.

In this paper the outage probability of a cognitive decode-and-forward relay network operating over Nakagami-m fading channels is evaluated. Based on the underlay approach, secondary transmissions are allowed only in cases where interference constraints on the primary destination receivers are satisfied. Additionally, by taking into consideration the interfering effects coming from the multiple primary transmitters, the signal-to-interference-plus-noise-ratio statistic of the secondary nodes is investigated. The derived results include exact expressions as well as approximated ones for high values of the maximum allowed transmitted power at the cognitive network and/or the interference limited case. We present numerical performance evaluation results for various channel conditions and communication scenarios. These results are complemented by equivalent computer simulated ones, which validate the accuracy of the proposed analysis.

In this paper, we study the problem of resource allocation for a multiuser orthogonal frequency-division multiple access (OFDMA) downlink with eavesdropping. The considered setup consists of a base station, several users and a single eavesdropper that indents to wiretap the transmitted message within each OFDMA subchannel. By taking into consideration the existence of the eavesdropper, the base station aims to assign subchannels and allocate the available power in order to optimize the max-min fairness criterion over the users' secrecy rate. The investigated problem is hard to be solved because of its combinatorial and nonlinear nature. Thus, optimality is discussed for two particular cases where the available power tends to infinity and zero, respectively. The optimal solution is obtained by formulating a mixed integer linear problem in the first case and a series of linear sum assignment problems in the second. In addition, two low-complexity solutions are presented which are based on decoupling the subchannel and the power allocation subproblems. Numerical results are provided to illustrate the performance of the presented solutions.

This paper addresses the problem of noise removal in X-ray medical images. A novel scheme for image denoising is proposed, by leveraging recent advances in sparse and redundant representations. The noisy X-ray image is decomposed, with respect to an overcomplete dictionary which is either fixed or trained on the noisy image, and it is reconstructed using greedy techniques. The new scheme has been tested with both artificial and real X-ray images and it turns out that it may offer superior denoising results as compared to other existing methods.

We introduce an adaptive distributed technique that is suitable for node-specific parameter estimation in an adaptive network where each node is interested in a set of parameters of local interest as well as a set of network global parameters. The estimation of each set of parameters of local interest is undertaken by a local Least Mean Squares (LMS) algorithm at each node. At the same time and coupled with the previous local estimation processes, an incremental mode of cooperation is implemented at all nodes in order to perform an LMS algorithm which estimates the parameters of global interest. In the steady state, the new distributed technique converges to the MMSE solution of a centralized processor that is able to process all the observations. To illustrate the effectiveness of the proposed technique we provide simulation results in the context of cooperative spectrum sensing in cognitive radio networks.

In this paper adaptive Eigenvalue-Based Spectrum Sensing (EBSS) techniques are proposed for multi-antenna cognitive receivers. In cases where fading channels are involved, and hence real-time processing is required, the adaptive techniques offer lower complexity and exhibit improved performance as compared to the corresponding batch EBSS techniques. At first, novel adaptive EBSS techniques are developed for the Maximum Eigenvalue Detector (MED), the Maximum-Minimum Eigenvalue Detector (MMED), and the Generalized Likelihood Ratio Test (GLRT) schemes, respectively, based on well-studied subspace tracking methods. Moreover, close approximations for the distribution functions of the adaptive test statistics of the MED, MMED and GLRT schemes are derived in order to compute the decision thresholds for a given probability of false alarm. The performance of the adaptive EBSS methods is verified via indicative simulations.

In this paper a new Opportunistic Interference Alignment (OIA) technique is presented that can be applied in a blind manner to Cognitive Radio systems. The proposed technique aims at relaxing the strong assumptions made in existing approaches where it is assumed that the Secondary User (SU) has knowledge of Channel State Information (CSI) concerning the Primary User's (PU) transmissions, although this may not be possible, in general. First, the OIA technique of [1] is extended so that the required CSI can be blindly estimated from the second order statistics of the received symbols. More specifically, the Secondary Receiver post-coding matrix is re-designed in order to null-out completely the interference generated by the PU's transmissions, in contrast to the original work where an interference reduction approach was adopted. Then, a novel estimation mechanism is proposed that enables the SU to apply the new OIA technique in a blind manner. The performance of the proposed technique is evaluated both theoretically as well as via simulations, under perfect and imperfect CSI.

In this paper, two new power policies (PPs), namely the Rate Maximization Policy (RMP) and the Rate-Bound Maximization Policy (RBMP), for Cognitive Radio (CR) systems are developed based on an expected rate maximization criterion. Both policies are designed such as to satisfy a constraint related to the average interference caused by the CR network to primary communication. The key novelty here is that the proposed PPs take into account the existence of limited (hybrid) channel feedback, where the direct secondary channel is known instantaneously while the interference caused by the primary transmission to the secondary receiver is only known statistically. The optimal policy (RMP) is characterized, and a low complexity algorithm is presented that allows for its efficient and accurate implementation. The two policies are compared and RMP is shown to lead to substantial energy consumption savings at equal rate performance.

In the present paper, we propose two novel algorithms which enable the relay cooperation for the distributed computation of the beamforming weights in a blind and adaptive manner, without the need to forward the data to a fusion center. In the first scheme, the beamforming vector is computed through minimization of the total transmit power subject to a receiver quality-of-service constraint (QoS). In the second scheme, the beamforming weights are obtained through maximization of the receiver signal-to-noise-ratio (SNR) subject to a total transmit power constraint. The proposed approaches distribute the computational overhead equally among the relay nodes and achieve close performance to the one of the optimal beamforming solutions. Note, that the aforementioned optimal solutions are derived assuming perfect channel state information at the relays' side. In order to verify the performance of the proposed approaches, indicative simulations were carried out for static and time-varying channels.

In this paper novel subspace-based blind schemes are proposed and applied to the sparse channel identification problem. Moreover, adaptive sparse subspace tracking methods are proposed so as to provide efficient real-time implementations. The new algorithms exploit the subspace sparsity either via employing ℓ1-norm relaxation or through greedy-based optimization. The derived schemes have been tested in a Zero-Prefix Orthogonal Frequency Division Multiplexing (ZP-OFDM) system and it turns out that, compared to state-of-art existing schemes, they offer improved performance in terms of convergence rate and steady-state error.

In this paper new efficient decision feedback equalization (DFE) schemes for channels with long and sparse impulse responses are proposed. It has been shown that under reasonable assumptions concerning the channel impulse response (CIR) coefficients, the feedforward (FF) and feedback (FB) filters may be also approximated by sparse filters. Either the sparsity of the CIR, or the sparsity of the DFE filters may be exploited to derive efficient implementations of the DFE. To this end, compressed sampling (CS) approaches, already successful in system identification settings, can significantly improve the performance of the non sparsity aware DFE. Building on basis pursuit and matching pursuit techniques new DFE schemes are proposed that exhibit considerable computational savings, increased performance properties and short training sequence requirements. To investigate the performance of the proposed schemes the restricted isometry property in the common DFE setup is also investigated.

While Half-Duplex cooperative systems provide an alternative means to achieve diversity gain, they suffer from multiplexing gain loss due to the necessary two-phase transmission protocols that are employed. In this paper two new transmission techniques are presented for systems with multi-antenna relay nodes employing the Decode-and-Forward (DF) protocol which achieve improved multiplexing and diversity gains. The first technique extends to the multi-antenna relays case our previous work for single-antenna relay nodes [5]-[6] and is based on the channel's matrix singular value decomposition (SVD). The second technique exploits the multiple-antenna relays so as to achieve Full-Duplex communication by permitting the latter ones to transmit and receive simultaneously. The transmissions' orthogonality is achieved via a novel distributed interference nulling mechanism. The proposed schemes' performance is theoretically studied and the results are verified through typical simulations.

In this paper, we study full-duplex (FD) operation from a diversity perspective and investigate several protocols that extract diversity gains over a block fading channel. The investigated approach introduces a block-by-block transmission and requires data codewords that span several independent realizations of channel fading. This fundamental approach ensures a diversity gain at least equal to one (i.e., it does not suffer from error floors) and is employed to different relaying strategies without loop interference cancellation (LIC) and with imperfect LIC: a) Amplify-and-Forward (AF) without LIC, b) AF with imperfect LIC and c) Decode-and-Forward with imperfect LIC. The proposed protocols are analyzed from a diversity-multiplexing tradeoff (DMT) standpoint and practical universal codes that achieve the DMT of the proposed schemes are also presented. We show that AF without LIC ensures a full time diversity independently on the statistics of the loop interference, and it is introduced as a general FD-based relaying scheme. In addition, we demonstrate that imperfect LIC is not efficient for cases with strong residual loop interference and limits the diversity gain to one.

In this paper, training design is studied for a single relay amplify-and-forward cooperative network. The taps of the frequency selective channels are assumed to be correlated and OFDM modulation is used for transmission. Based on the least squares (LS) criterion, conditions for pilot tone positioning and a number of power allocation schemes are described. Analytical closed-form power allocation expressions are provided for both the source and the relay. The theoretical results are fully corroborated by simulations.

An analytical bit error rate (BER) performance study of three detect-and-forward (DaF) policies under a common framework is presented in this paper. More specifically, the direct, thresholded and link-adaptive schemes are studied, which differ in the way the decision on the transmitted symbol is forwarded from the relay to the destination. The analysis is carried out for a single relay DaF transmission protocol and takes into account the symbol decision errors that may occur at the relay. Simple closed-form analytical BER expressions are derived for all three schemes, which are completely verified by simulations. Moreover, by minimizing a properly selected conditional BER expression, a novel optimal DaF strategy is proposed, which outperforms all previously known related methods.

A new technique for the uplink transmission of a cooperative system consisting of single-antenna source and relay nodes and a multi-antenna destination node is presented. The proposed technique is based on the singular value decomposition of the channel matrix and, for the same rate, it exhibits higher diversity gain as compared to the existing ones. Theoretical analysis of the technique is carried out when the Decode-and-Forward protocol is employed. The analysis reveals the way the decoding errors occurring at the relays degrade the performance of the system. Furthermore, the proposed technique is extended to a multi-user environment so as to exploit multi-user diversity. A novel criterion is suggested for selecting the best transmitting user at each time slot. The performance of the system in the multi-user case is theoretically studied and the number of users needed so as the system to achieve its maximum performance is computed. The derived theoretical results are verified via typical simulations.

The performance of dual-hop Decode-and-Forward relaying with relay selection (RS) and repetitive transmission is studied over Nakagami-m fading channels. Closed-form expressions for the moment generating function of RS-based transmission with maximal-ratio diversity at the destination and repetitive transmission with selection diversity (SD) are derived. We obtain the outage probability (OP) and the average symbol error probability (ASEP) for the following relaying schemes: repetitive transmission, pure RS-based transmission and a rate-selective scheme that utilizes the selected relay node only if it provides higher achievable rate than direct transmission. Performance results show that RS-based schemes always outperform the repetitive ones. Furthermore it is shown that in terms of ASEP pure RS is always beneficial, although in terms of OP the single-hop transmission outperforms dual-hop relaying if the source to destination channel is sufficiently strong.

The performance of dual-hop Decode-and-Forward relaying with relay selection (RS) is analyzed over Nakagami-$m$fading channels. Assuming that the direct source-to-destination link is active, closed-form expressions for the moment generating and the cumulative distribution functions of a RS-based cooperation scheme that utilizes maximal-ratio diversity at the destination are derived. These expressions are used to obtain the outage probability (OP) and average symbol error probability (ASEP) of this pure RS scheme as well as of a rate-selective one that utilizes RS only when it provides higher achievable rate than that of the direct transmission. Numerically evaluated results, verified by computer simulations, show that, although, in terms of ASEP, relaying is always beneficial, in terms of OP, it should be disabled whenever the direct link is strong.

A unified analytical framework for evaluating the performance of maximal ratio combining (MRC) and orthogonal space-time block coding (OSTBC) over generalized fading channels is presented. The basic motivation for developing such a framework pertains to analyzing the distribution of a sum of squared random variables (RVs) belonging to different families of fading distributions. Following a novel analytical approach stemming from the definition of a common moment generating function (MGF) model for these families of distributions, the probability density function (PDF) and the cumulative distribution function (CDF) of a general sum of squared RVs are expressed by simple infinite Gamma series expansions. Based on these expressions, the capacity and error probability of MRC/OSTBC over generalized fading channels is thoroughly studied. The developed theory is used to evaluate the performance of OSTBC for a mixed Nakagami-m/Rice fading model and novel analytical results are presented. The accuracy of the derived expressions is validated by extensive Monte Carlo simulations.

New results for the multichannel Nakagami-m fading model with an arbitrary correlation matrix are presented in this paper. By using an efficient tridiagonalization method based on Householder matrices, the inverse of the Gaussian correlation matrix is transformed to tridiagonal, managing to derive a closed-form union upper bound for the joint Nakagami-m probability density function and an exact analytical expression for the moment generating function of the sum of identically distributed gamma random variables. Our analysis considers an arbitrary correlation structure, which includes as special cases the exponential, constant, circular, and linear correlation ones. Based on the proposed mathematical analysis, we obtain a tight union upper bound for the outage probability of multibranch selection diversity receivers as well as exact analytical expressions for the outage and the average error probability of multibranch maximal-ratio diversity receivers. Our analysis is verified by comparing numerically evaluated with extensive computer simulation performance evaluation results, showing the usefulness of the proposed approach.

New results for the multichannel Nakagami-m fading model with an arbitrary correlation matrix are presented. By using an efficient tridiagonalization method based on Householder matrices, a union upper bound for the joint Nakagami-m probability density function and an infinite series representation for the moment generating function of the sum of gamma random variables are derived. Based on the proposed mathematical analysis, a tight union upper bound for the outage probability of multibranch selection diversity, as well as, exact analytical expressions for the outage and average error probability of multibranch maximal-ratio diversity receivers operating over identically distributed and arbitrarily correlated Nakagami-m fading channels are obtained. Our analysis is verified by comparisons of numerically evaluated results with extensive computer simulation ones.

The equalization of MIMO frequency selective channels is a challenging task due to the increased filter lengths required and the presence of interstream interference. In wideband systems where high delay spread channels arise and long transmission bursts are employed, adaptive equalization is necessary. In this work, we develop an adaptive MIMO DFE for wideband time varying systems. To reduce computational complexity and accelerate convergence rate, the equalizer is fully implemented in the frequency domain, while an inherent causality problem is efficiently solved through an iterative scheme. As verified by simulations, the proposed method offers a good trade off in terms of complexity and convergence, compared to other related adaptive algorithms.

In this paper, new results for the multivariate Rayleigh distribution with an arbitrary correlation matrix are presented. By using an efficient tridiagonalization method based on Householder matrices, closed-form union upper bounds for its joint probability density and cumulative density functions are derived. Based on the proposed mathematical analysis, a tight union upper bound and an analytical approximation for the outage probability of multibranch selection diversity receivers operating over identically distributed and arbitrarily correlated Rayleigh fading channels are obtained. Our analysis is verified by comparing numerically evaluated results with extensive computer simulation ones.

An infinite series representation for the moment generating function of the sum of squared arbitrarily correlated Rayleigh random variables is presented. Based on the derived formula, corresponding analytical expressions for the probability density and cumulative distribution functions are extracted. As an application for the aforementioned sum, exact analytical expressions for the outage and the average error probability, as well as, the channel average spectral efficiency of multibranch maximal-ratio diversity receivers operating over identically distributed and arbitrarily correlated Rayleigh fading channels are obtained. Our analysis is verified by comparing numerically evaluated results with extensive computer simulation ones.

Three new adaptive equalization algorithms for wireless systems operating over frequency selective MIMO channels are proposed. The problem of the MIMO DFE design is formulated as a set of linear equations with multiple right-hand sides (RHS) evolving in time. By applying an adaptive modified conjugate gradient algorithm, originally proposed for a single linear system, to the problem at hand, we arrive at an equalizer of performance identical to RLS, numerically robust, but of higher computational cost. To reduce its complexity, two updating strategies of the equalizer filters are derived based on Galerkin projection in time and space respectively. The two alternative schemes exhibit a complexity lower than RLS while offering slightly inferior convergence properties.

Among the methods that have been proposed for a Multiple-Input Multiple-Output (MIMO) receiver, the V-BLAST algorithm provides a good compromise between transmission rate, achievable diversity, and decoding complexity. In this paper, we derive a new adaptive V-BLAST type equalization scheme for fast time varying, flat fading MIMO channels. The proposed equalizer stems from the Cholesky factorization of MIMO system's output data autocorrelation matrix and the equalizer's filters are updated in time using numerically robust unitary Givens rotations. The new square-root algorithm exhibits identical performance to a recently proposed V-BLAST adaptive algorithm, offering at the same time substantially reduced computational complexity. Moreover, as expected due to its square-root form and verified by simulations, the algorithm exhibits particularly favourable numerical behaviour.

Among the methods that have been proposed for a multiple-input multiple-output (MIMO) receiver, the V-BLAST algorithm provides a good compromise between transmission rate, achievable diversity, and decoding complexity. In this letter, we derive a new adaptive V-BLAST-type equalization scheme for fast time-varying, flat-fading MIMO channels. The proposed equalizer stems from the Cholesky factorization of the MIMO system's output data autocorrelation matrix, and the equalizer filters are updated in time using numerically robust unitary Givens rotations. The new square-root algorithm exhibits identical performance to a recently proposed V-BLAST adaptive algorithm, offering at the same time noticeable reduction in computational complexity. Moreover, as expected due to its square-root form and verified by simulations, the algorithm exhibits particularly favorable numerical behavior.

The aim of this paper is to improve the performance of spatial domain watermarking. To this end, a new perceptual mask and a new detection scheme are proposed. The proposed spatial perceptual mask is based on the cover image prediction error sequence and matches very well with the properties of the human visual system. It exhibits superior performance compared to existing spatial masking schemes. Moreover, it allows for a significantly increased strength of the watermark while, at the same time, the watermark visibility is decreased. The new blind detection scheme comprises an efficient prewhitening process and a correlation-based detector. The prewhitening process is based on the least-squares prediction error filter and substantially improves the detector's performance. The correlation-based detector that was selected is shown to be the most suitable for the problem at hand. The improved performance of the proposed detection scheme has been justified theoretically for the case of linear filtering plus noise attack and through extensive simulations. The theoretical analysis is independent of the proposed mask and the derived expressions can be used for any watermarking technique based on spatial masking. It is shown though that in most cases the detector performs better if the proposed mask is employed.

A parametric method for estimating the unknown channel impulse response (CIR) in a semi-blind manner is proposed. The main trait of this method is that, instead of seeking the whole CIR sequence, only the unknown time delays and attenuation factors of the physical channel multipath components are estimated. The technique is based on a suitable application of the subchannel response matching (SRM) criterion. The resulting cost function is separable with respect to the two sets of unknown parameters, i.e. time delays and attenuations. Thus, an efficient two step optimization procedure can be applied. The new method offers significant computational savings and a lower mean square estimation error as compared to existing semi-blind channel estimation methods.

A new efficient decision feedback equalizer (DFE) appropriate for channels with long and sparse impulse response (IR) is proposed. Such channels are encountered in many high-speed wireless communications applications. It is shown that, in cases of sparse channels, the feedforward and feedback (FB) filters of the DFE have a particular structure, which can be exploited to derive efficient implementations of the DFE, provided that the time delays of the channel IR multipath components are known. This latter task is accomplished by a novel technique, which estimates the time delays based on the form of the channel input-output cross-correlation sequence in the frequency domain. A distinct feature of the resulting DFE is that the involved FB filter consists of a reduced number of active taps. As a result, it exhibits considerable computational savings, faster convergence, and improved tracking capabilities as compared with the conventional DFE. Note that faster convergence implies that a shorter training sequence is required. Moreover, the new algorithm has a simple form and its steady-state performance is almost identical to that of the conventional DFE.

An adaptive equalization scheme based on channel estimation is proposed. The structure of the whole scheme is novel as well as some of its most important constituent parts. The equalization part comprises an adaptive decision feedback equalizer (DFE) whose performance is significantly improved due to the information provided by the channel estimation part. To estimate the channel impulse response (CIR) a new adaptive semi-blind scheme based on the sub-channel response matching (SRM) criterion and implemented in the frequency domain has been developed. The equalization part exhibits a much faster convergence rate as compared to the conventional DFE, hence a much shorter training sequence is required. Moreover a significant reduction of the error propagation phenomenon can be achieved. The above advantages are offered with only a slight increase in computational load compared to classical LMS-based DFE.

Two new adaptive equalizers are proposed which belong to the quasi-Newton (QN) algorithmic family. The first algorithm. is a linear equalizer (LE) and the second one is a decision feedback equalizer (DFE). In the LE case, the involved inverse Hessian matrix is approximated by a proper expansion consisting of powers of a Toeplitz matrix. Due to this formulation, the algorithm can be efficiently implemented in the transform domain (TD) using FFT. The same idea is applied to the feedforward part of the DFE. The derived algorithms enjoy the advantages of QN algorithms, that is, they exhibit faster convergence than their stochastic gradient counterparts and less computational complexity as compared to other Newton-type algorithms. These advantages are further enhanced due to TD implementation.

In this paper a computationally e cient Decision Feedback Equalizer (DFE) is proposed. The new equalizer is appropriate for channels with long and sparse impulse response (IR) as those encountered in many wireless communications applications. The main feature of the algorithm is that the actual size of the computationally demanding feedback lter is signi cantly reduced. This is achieved by exploiting the particular form of the multipath channel to derive a tractable expression for the causal part of the overall discrete channel IR (including the feedforward lter). Based on the above expression the feedback lter can be built so as to act only to a properly selected set of tap positions. The new DFE exhibits considerable computational savings and faster convergence as compared to the conventional DFE, o ering the same steady-state performance. 1.

A new block adaptive decision feedback equalizer is developed. Both the feedforward (FF) and the feedback (FB) filters are updated once every K sample time intervals, with K being the block length. It should be noted that this block adaptation is done in such a way that the resulting filters, and the decisions as well, are identical to those computed by the conventional sample by sample LMS-based decision feedback equalizer (LMS-DFE). The new algorithm offers substantial computational savings as compared to the sample-by-sample LMS-DFE with no loss in performance. The new block DFE turns out to be particularly suitable for applications requiring long equalizers.

A new block adaptive filtering algorithm is derived. The algorithm belongs to the quasi-Newton class of algorithms which are known to have superior convergence properties as compared to gradient algorithms. An important feature of the algorithm is that it lends itself to efficient implementation in the frequency domain. The computational complexity of the frequency domain version of the algorithm is comparable to that of the self-orthogonalizing frequency domain block LMS algorithm while it converges at a faster rate. Depending on the input signal this improvement in performance may be significant in some practical cases.

The algorithm presented in the paper is an exact block processing counterpart of the fast Newton transversal filtering (FNTF) algorithm [Moustakides and Theodorides, 1991]. The main trait of the new algorithm is that the block processing is done in such a way so that the resulting estimates are mathematically equivalent with the respective estimates of the FNTF algorithm. In cases where the involved filter is of medium to long order the new algorithm offers a substantial saving in computational complexity without sacrificing performance.