In ill-posed imaging inverse problems, there can exist many hypotheses that fit both the observed measurements and prior knowledge of the true image. Rather than returning just one hypothesis of that image, posterior samplers aim to explore the full solution space by generating many probable hypotheses, which can later be used to quantify uncertainty or construct recoveries that appropriately navigate the perception/distortion trade-off. In this work, we propose a fast and accurate posterior-sampling conditional generative adversarial network (cGAN) that, through a novel form of regularization, aims for correctness in the posterior mean as well as the trace and K principal components of the posterior covariance matrix. Numerical experiments demonstrate that our method outperforms competitors in a wide range of ill-posed imaging inverse problems.

Accelerated magnetic resonance (MR) imaging attempts to reduce acquisition time by collecting data below the Nyquist rate. As an ill-posed inverse problem, many plausible solutions exist, yet the majority of deep learning approaches generate only a single solution. We instead focus on sampling from the posterior distribution, which provides more comprehensive information for downstream inference tasks. To do this, we design a novel conditional normalizing flow (CNF) that infers the signal component in the measurement operator’s nullspace, which is later combined with measured data to form complete images. Using fastMRI brain and knee data, we demonstrate fast inference and accuracy that surpasses recent posterior sampling techniques for MRI. Code is available at https://github.com/jwen307/mri_cnf

In image recovery problems, one seeks to infer an image from distorted, incomplete, and/or noise-corrupted measurements. Such problems arise in magnetic resonance imaging (MRI), computed tomography, deblurring, super-resolution, inpainting, phase retrieval, image-to-image translation, and other applications. Given a training set of signal/measurement pairs, we seek to do more than just produce one good image estimate. Rather, we aim to rapidly and accurately sample from the posterior distribution. To do this, we propose a regularized conditional Wasserstein GAN that generates dozens of high-quality posterior samples per second. Our regularization comprises an $\ell_1$ penalty and an adaptively weighted standard-deviation reward. Using quantitative evaluation metrics like conditional Fréchet inception distance, we demonstrate that our method produces state-of-the-art posterior samples in both multicoil MRI and large-scale inpainting applications. The code for our model can be found here: https://github.com/matt-bendel/rcGAN.

Plug-and-play (PnP) methods that employ application-specific denoisers have been proposed to solve inverse problems, including MRI reconstruction. However, training application-specific denoisers is not feasible for many applications due to the lack of training data. In this work, we propose a PnP-inspired recovery method that does not require data beyond the single, incomplete set of measurements. The proposed self-supervised method, called recovery with a self-calibrated denoiser (ReSiDe), trains the denoiser from the patches of the image being recovered. The denoiser training and a call to the denoising subroutine are performed in each iteration of a PnP algorithm, leading to a progressive refinement of the reconstructed image. For validation, we compare ReSiDe with a compressed sensing-based method and a PnP method with BM3D denoising using single-coil MRI brain data.

For image recovery problems, plug-and-play (PnP) methods have been developed that replace the proximal step in an optimization algorithm with a call to an application-specific denoiser, often implemented using a deep neural network. Although such methods have been successful, they can be improved. For example, the denoiser is often trained using white Gaussian noise, while PnP's denoiser input error is often far from white and Gaussian, with statistics that are difficult to predict from iteration to iteration. PnP methods based on approximate message passing (AMP) are an exception, but only when the forward operator behaves like a large random matrix. In this work, we design a PnP method using the expectation consistent (EC) approximation algorithm, a generalization of AMP, that offers predictable error statistics at each iteration, from which a deep-net denoiser can be effectively trained.

Motivated by image recovery in magnetic resonance imaging (MRI), we propose a new approach to solving linear inverse problems based on iteratively calling a deep neural-network, sometimes referred to as plug-and-play recovery. Our approach is based on the vector approximate message passing (VAMP) algorithm, which is known for mean-squared error (MSE)-optimal recovery under certain conditions. The forward operator in MRI, however, does not satisfy these conditions, and thus we design new damping and initialization schemes to help VAMP. The resulting DD-VAMP++ algorithm is shown to outperform existing algorithms in convergence speed and accuracy when recovering images from the fastMRI database for the practical case of Cartesian sampling.

We consider the linear regression problem, where the goal is to recover the vector x ∈ ℝnfrom measurements y = Ax+w ∈ ℝmunder known matrix A and unknown noise w. For large i.i.d. sub-Gaussian A, the approximate message passing (AMP) algorithm is precisely analyzable through a state-evolution (SE) formalism, which furthermore shows that AMP is Bayes optimal in certain regimes. The rigorous SE proof, however, is long and complicated. And, although the AMP algorithm can be derived as an approximation of loop belief propagation (LBP), this viewpoint provides little insight into why large i.i.d. A matrices are important for AMP, and why AMP has a state evolution. In this work, we provide a heuristic derivation of AMP and its state evolution, based on the idea of "firstorder cancellation," that provides insights missing from the LBP derivation while being much shorter than the rigorous SE proof.

We consider the problem of estimating the input and hidden variables of a stochastic multi-layer neural network from an observation of the output. The hidden variables in each layer are represented as matrices with statistical interactions along both rows as well as columns. This problem applies to matrix imputation, signal recovery via deep generative prior models, multi-task and mixed regression, and learning certain classes of two-layer neural networks. We extend a recently-developed algorithm -- Multi-Layer Vector Approximate Message Passing (ML-VAMP), for this matrix-valued inference problem. It is shown that the performance of the proposed Multi-Layer Matrix VAMP (ML-Mat-VAMP) algorithm can be exactly predicted in a certain random large-system limit, where the dimensions $N\times d$ of the unknown quantities grow as $N\rightarrow\infty$ with $d$ fixed. In the two-layer neural-network learning problem, this scaling corresponds to the case where the number of input features, as well as training samples, grow to infinity but the number of hidden nodes stays fixed. The analysis enables a precise prediction of the parameter and test error of the learning.

We consider the non-coherent single-input multiple-output (SIMO) multiple access channel with general signaling under spatially correlated Rayleigh block fading. We propose a novel soft-output multi-user detector that computes an approximate marginal posterior of each transmitted signal using only the knowledge about the channel distribution. Our detector is based on expectation propagation (EP) approximate inference and has polynomial complexity in the number of users, number of receive antennas and channel coherence time. We also propose two simplifications of this detector with reduced complexity. With Grassmannian signaling, the proposed detectors outperform a state-of-the-art non-coherent detector with projection-based interference mitigation. With pilot-assisted signaling, the EP detector outperforms, in terms of symbol error rate, some conventional coherent pilot-based detectors, including a sphere decoder and a joint channel estimation-data detection scheme. Our EP-based detectors produce accurate approximates of the true posterior leading to high achievable sum-rates. The gains of these detectors are further observed in terms of the bit error rate when using their soft outputs for a turbo channel decoder.

The standard linear regression (SLR) problem is to recover a vector x0from noisy linear observations y = Ax0+ w. The approximate message passing (AMP) algorithm proposed by Donoho, Maleki, and Montanari is a computationally efficient iterative approach to SLR that has a remarkable property: for large i.i.d. sub-Gaussian matrices A, its per-iteration behavior is rigorously characterized by a scalar state-evolution whose fixed points, when unique, are Bayes optimal. The AMP algorithm, however, is fragile in that even small deviations from the i.i.d. sub-Gaussian model can cause the algorithm to diverge. This paper considers a “vector AMP” (VAMP) algorithm and shows that VAMP has a rigorous scalar state-evolution that holds under a much broader class of large random matrices A: those that are right-orthogonally invariant. After performing an initial singular value decomposition (SVD) of A, the per-iteration complexity of VAMP is similar to that of AMP. In addition, the fixed points of VAMP's state evolution are consistent with the replica prediction of the minimum mean-squared error derived by Tulino, Caire, Verdú, and Shamai. Numerical experiments are used to confirm the effectiveness of VAMP and its consistency with state-evolution predictions.

Phase retrieval algorithms have become an important component in many modern computational imaging systems. For instance, in the context of ptychography and speckle correlation imaging, they enable imaging past the diffraction limit and through scattering media, respectively. Unfortunately, traditional phase retrieval algorithms struggle in the presence of noise. Progress has been made recently on developing more robust algorithms using signal priors, but at the expense of limiting the range of supported measurement models (e.g., to Gaussian or coded diffraction patterns). In this work we leverage the regularization-by-denoising framework and a convolutional neural network denoiser to create prDeep, a new phase retrieval algorithm that is both robust and broadly applicable. We test and validate prDeep in simulation to demonstrate that it is robust to noise and can handle a variety of system models.

Deep generative networks provide a powerful tool for modeling complex data in a wide range of applications. In inverse problems that use these networks as generative priors on data, one must often perform inference of the inputs of the networks from the outputs. Inference is also required for sampling during stochastic training of these generative models. This paper considers inference in a deep stochastic neural network where the parameters (e.g., weights, biases and activation functions) are known and the problem is to estimate the values of the input and hidden units from the output. A novel and computationally tractable inference method called Multi-Layer Vector Approximate Message Passing (ML-VAMP) is presented. Our main contribution shows that the mean-squared error (MSE) of ML-VAMP can be exactly predicted in a certain large system limit. In addition, the MSE achieved by ML-VAMP matches the Bayes optimal value recently postulated by Reeves when certain fixed point equations have unique solutions.

Estimating a vector $\mathbf{x}$ from noisy linear measurements $\mathbf{Ax+w}$ often requires use of prior knowledge or structural constraints on $\mathbf{x}$ for accurate reconstruction. Several recent works have considered combining linear least-squares estimation with a generic or plug-in ``denoiser" function that can be designed in a modular manner based on the prior knowledge about $\mathbf{x}$. While these methods have shown excellent performance, it has been difficult to obtain rigorous performance guarantees. This work considers plug-in denoising combined with the recently-developed Vector Approximate Message Passing (VAMP) algorithm, which is itself derived via Expectation Propagation techniques. It shown that the mean squared error of this ``plug-in" VAMP can be exactly predicted for a large class of high-dimensional random $\Abf$ and denoisers. The method is illustrated in image reconstruction and parametric bilinear estimation.

Vector approximate message passing (VAMP) is a computationally simple approach to the recovery of a signal x from noisy linear measurements y = Ax + w. Like the AMP proposed by Donoho, Maleki, and Montanari in 2009, VAMP is characterized by a rigorous state evolution (SE) that holds under certain large random matrices and that matches the replica prediction of optimality. But while AMP's SE holds only for large i.i.d. sub-Gaussian A, VAMP's SE holds under the much larger class: right-rotationally invariant A. To run VAMP, however, one must specify the statistical parameters of the signal and noise. This work combines VAMP with Expectation-Maximization to yield an algorithm, EM-VAMP, that can jointly recover x while learning those statistical parameters. The fixed points of the proposed EM-VAMP algorithm are shown to be stationary points of a certain constrained free-energy, providing a variational interpretation of the algorithm. Numerical simulations show that EM-VAMP is robust to highly ill-conditioned A with performance nearly matching oracle-parameter VAMP.

The standard linear regression (SLR) problem is to recover a vector x0from noisy linear observations y = Ax0+ w. The approximate message passing (AMP) algorithm recently proposed by Donoho, Maleki, and Montanari is a computationally efficient iterative approach to SLR that has a remarkable property: for large i.i.d. sub-Gaussian matrices A, its periteration behavior is rigorously characterized by a scalar stateevolution whose fixed points, when unique, are Bayes optimal. AMP, however, is fragile in that even small deviations from the i.i.d. sub-Gaussian model can cause the algorithm to diverge. This paper considers a “vector AMP” (VAMP) algorithm and shows that VAMP has a rigorous scalar state-evolution that holds under a much broader class of large random matrices A: those that are right-rotationally invariant. After performing an initial singular value decomposition (SVD) of A, the per-iteration complexity of VAMP is similar to that of AMP. In addition, the fixed points of VAMP's state evolution are consistent with the replica prediction of the minimum mean-squared error recently derived by Tulino, Caire, Verdú, and Shamai.

The problem of estimating a random vector x from noisy linear measurements y=Ax+w with unknown parameters on the distributions of x and w, which must also be learned, arises in a wide range of statistical learning and linear inverse problems. We show that a computationally simple iterative message-passing algorithm can provably obtain asymptotically consistent estimates in a certain high-dimensional large-system limit (LSL) under very general parameterizations. Previous message passing techniques have required i.i.d. sub-Gaussian A matrices and often fail when the matrix is ill-conditioned. The proposed algorithm, called adaptive vector approximate message passing (Adaptive VAMP) with auto-tuning, applies to all right-rotationally random A. Importantly, this class includes matrices with arbitrarily bad conditioning. We show that the parameter estimates and mean squared error (MSE) of x in each iteration converge to deterministic limits that can be precisely predicted by a simple set of state evolution (SE) equations. In addition, a simple testable condition is provided in which the MSE matches the Bayes-optimal value predicted by the replica method. The paper thus provides a computationally simple method with provable guarantees of optimality and consistency over a large class of linear inverse problems.

Generalized linear models, where a random vector x is observed through a noisy, possibly nonlinear, function of a linear transform z = Ax, arise in a range of applications in nonlinear filtering and regression. Approximate message passing (AMP) methods, based on loopy belief propagation, are a promising class of approaches for approximate inference in these models. AMP methods are computationally simple, general, and admit precise analyses with testable conditions for optimality for large i.i.d. transforms A. However, the algorithms can diverge for general A. This paper presents a convergent approach to the generalized AMP (GAMP) algorithm based on direct minimization of a large-system limit approximation of the Bethe free energy (LSL-BFE). The proposed method uses a double-loop procedure, where the outer loop successively linearizes the LSL-BFE and the inner loop minimizes the linearized LSL-BFE using the alternating direction method of multipliers (ADMM). The proposed method, called ADMM-GAMP, is similar in structure to the original GAMP method, but with an additional least-squares minimization. It is shown that for strictly convex, smooth penalties, ADMM-GAMP is guaranteed to converge to a local minimum of the LSL-BFE, thus providing a convergent alternative to GAMP that is stable under arbitrary transforms. Simulations are also presented that demonstrate the robustness of the method for non-convex penalties as well.

Approximations of loopy belief propagation, including expectation propagation and approximate message passing, have attracted considerable attention for probabilistic inference problems. This paper proposes and analyzes a generalization of Opper and Winther's expectation consistent (EC) approximate inference method. The proposed method, called Generalized Expectation Consistency (GEC), can be applied to both maximum a posteriori (MAP) and minimum mean squared error (MMSE) estimation. Here we characterize its fixed points, convergence, and performance relative to the replica prediction of optimality.

The estimation of a random vector with independent components passed through a linear transform followed by a componentwise (possibly nonlinear) output map arises in a range of applications. Approximate message passing (AMP) methods, based on Gaussian approximations of loopy belief propagation, have recently attracted considerable attention for such problems. For large random transforms, these methods exhibit fast convergence and admit precise analytic characterizations with testable conditions for optimality, even for certain non-convex problem instances. However, the behavior of AMP under general transforms is not fully understood. In this paper, we consider the generalized AMP (GAMP) algorithm and relate the method to more common optimization techniques. This analysis enables a precise characterization of the GAMP algorithm fixed points that applies to arbitrary transforms. In particular, we show that the fixed points of the so-called max-sum GAMP algorithm for MAP estimation are critical points of a constrained maximization of the posterior density. The fixed points of the sum-product GAMP algorithm for estimation of the posterior marginals can be interpreted as critical points of a certain free energy.

In cosparse analysis compressive sensing (CS), one seeks to estimate a non-sparse signal vector from noisy sub-Nyquist linear measurements by exploiting the knowledge that a given linear transform of the signal is cosparse, i.e., has sufficiently many zeros. We propose a novel approach to cosparse analysis CS based on the generalized approximate message passing (GAMP) algorithm. Unlike other AMP-based approaches to this problem, ours works with a wide range of analysis operators and regularizers. In addition, we propose a novel ℓ0-like soft-thresholder based on MMSE denoising for a spike-and-slab distribution with an infinite-variance slab. Numerical demonstrations on synthetic and practical datasets demonstrate advantages over existing AMP-based, greedy, and reweighted-ℓ1approaches.

The generalized approximate message passing (GAMP) algorithm is an efficient method of MAP or approximate-MMSE estimation of x observed from a noisy version of the transform coefficients z = Ax. In fact, for large zero-mean i.i.d sub-Gaussian A, GAMP is characterized by a state evolution whose fixed points, when unique, are optimal. For generic A, however, GAMP may diverge. In this paper, we propose adaptive-damping and mean-removal strategies that aim to prevent divergence. Numerical results demonstrate significantly enhanced robustness to non-zero-mean, rank-deficient, column-correlated, and ill-conditioned A.

Generalized Linear Models (GLMs), where a random vector x is observed through a noisy, possibly nonlinear, function of a linear transform z = Ax arise in a range of applications in nonlinear filtering and regression. Approximate Message Passing (AMP) methods, based on loopy belief propagation, are a promising class of approaches for approximate inference in these models. AMP methods are computationally simple, general, and admit precise analyses with testable conditions for optimality for large i.i.d. transforms A. However, the algorithms can easily diverge for general transforms. This paper presents a convergent approach to the generalized AMP (GAMP) algorithm based on direct minimization of a large-system limit approximation of the Bethe Free Energy (LSL-BFE). The proposed method uses a double-loop procedure, where the outer loop successively linearizes the LSL-BFE and the inner loop minimizes the linearized LSL-BFE using the Alternating Direction Method of Multipliers (ADMM). The proposed method, called ADMM-GAMP, is similar in structure to the original GAMP method, but with an additional least-squares minimization. It is shown that for strictly convex, smooth penalties, ADMM-GAMP is guaranteed to converge to a local minima of the LSL-BFE, thus providing a convergent alternative to GAMP that is stable under arbitrary transforms. Simulations are also presented that demonstrate the robustness of the method for non-convex penalties as well.

Approximate message passing (AMP) methods and their variants have attracted considerable recent attention for the problem of estimating a random vector x observed through a linear transform A. In the case of large i.i.d. A, the methods exhibit fast convergence with precise analytic characterizations on the algorithm behavior. However, the convergence of AMP under general transforms is not fully understood. In this paper, we provide sufficient conditions for the convergence of a damped version of the generalized AMP (GAMP) algorithm in the case of Gaussian distributions. It is shown that, with sufficient damping the algorithm can be guaranteed to converge, but the amount of damping grows with peak-to-average ratio of the squared singular values of A. This condition explains the good performance of AMP methods on i.i.d. matrices, but also their difficulties with other classes of transforms. A related sufficient condition is then derived for the local stability of the damped GAMP method under more general (possibly non-Gaussian) distributions, assuming certain strict convexity conditions.

The articles in this special issue focus on the technology and applications supported by in-band full duplex wireless services.

In-band full-duplex (IBFD) operation has emerged as an attractive solution for increasing the throughput of wireless communication systems and networks. With IBFD, a wireless terminal is allowed to transmit and receive simultaneously in the same frequency band. This tutorial paper reviews the main concepts of IBFD wireless. One of the biggest practical impediments to IBFD operation is the presence of self-interference, i.e., the interference that the modem's transmitter causes to its own receiver. This tutorial surveys a wide range of IBFD self-interference mitigation techniques. Also discussed are numerous other research challenges and opportunities in the design and analysis of IBFD wireless systems.

The estimation of a random vector with independent components passed through a linear transform followed by a componentwise (possibly nonlinear) output map arises in a range of applications. Approximate message passing (AMP) methods, based on Gaussian approximations of loopy belief propagation, have recently attracted considerable attention for such problems. For large random transforms, these methods exhibit fast convergence and admit precise analytic characterizations with testable conditions for optimality, even for certain non-convex problem instances. However, the behavior of AMP under general transforms is not fully understood. In this paper, we consider the generalized AMP (GAMP) algorithm and relate the method to more common optimization techniques. This analysis enables a precise characterization of the GAMP algorithm fixed-points that applies to arbitrary transforms. In particular, we show that the fixed points of the so-called max-sum GAMP algorithm for MAP estimation are critical points of a constrained maximization of the posterior density. The fixed-points of the sum-product GAMP algorithm for estimation of the posterior marginals can be interpreted as critical points of a certain mean-field variational optimization.

In this paper, we consider the downlink of large OFDMA-based networks and study their performance bounds as a function of the number of - transmitters B, users K, and resource-blocks N. Here, a resource block is a collection of subcarriers such that all such collections, that are disjoint have associated independently fading channels. In particular, we analyze the expected achievable sum-rate as a function of above variables and derive novel upper and lower bounds for a general spatial geometry of transmitters, a truncated path-gain model, and a variety of fading models. We establish the associated scaling laws for dense and extended networks, and propose design guidelines for the regulators to guarantee various QoS constraints and, at the same time, maximize revenue for the service providers. Thereafter, we develop a distributed resource allocation scheme that achieves the same sum-rate scaling as that of the proposed upper bound for a wide range of K, B, N. Based on it, we compare low-powered peer-to-peer networks to high-powered single-transmitter networks and give an additional design principle. Finally, we also show how our results can be extended to the scenario where each of the B transmitters have M (>;1) co-located antennas.

In this paper, we consider the downlink of large-scale multi-cellular OFDMA-based networks and study performance bounds of the system as a function of the number of users K, the number of base-stations B, and the number of resource-blocks N. Here, a resource block is a collection of subcarriers such that all such collections, that are disjoint have associated independently fading channels. We derive novel upper and lower bounds on the sum-utility for a general spatial geometry of base stations, a truncated path loss model, and a variety of fading models (Rayleigh, Nakagami-m, Weibull, and LogNormal). We also establish the associated scaling laws and show that, in the special case of fixed number of resource blocks, a grid-based network of base stations, and Rayleigh-fading channels, the sum information capacity of the system scales as Θ(B log log K/B) for extended networks, and as O(B log log K) and Ω(log log K) for dense networks. Interpreting these results, we develop some design principles for the service providers along with some guidelines for the regulators in order to achieve provisioning of various QoS guarantees for the end users and, at the same time, maximize revenue for the service providers.

Gaussian and quadratic approximations of message passing algorithms on graphs have attracted considerable attention due to their computational simplicity, analytic tractability, and wide applicability in optimization and statistical inference problems. This paper summarizes a systematic framework for incorporating such approximate message passing (AMP) methods in general graphical models. The key concept is a partition of dependencies of a general graphical model into strong and weak edges, with each weak edge representing a small, linearizable coupling of variables. AMP approximations based on the central limit theorem can be applied to the weak edges and integrated with standard message passing updates on the strong edges. The resulting algorithm, which we call hybrid generalized approximate message passing (Hybrid-GAMP), can yield significantly simpler implementations of sum-product and max-sum loopy belief propagation. By varying the partition between strong and weak edges, a performance-complexity trade-off can be achieved. Structured sparsity problems are studied as an example of this general methodology where there is a natural partition of edges.

In this paper we consider the problem of full-duplex multiple-input multiple-output (MIMO) relaying between multi-antenna source and destination nodes. The principal difficulty in implementing such a system is that, due to the limited attenuation between the relay's transmit and receive antenna arrays, the relay's outgoing signal may overwhelm its limited-dynamic-range input circuitry, making it difficult-if not impossible-to recover the desired incoming signal. While explicitly modeling transmitter/receiver dynamic-range limitations and channel estimation error, we derive tight upper and lower bounds on the end-to-end achievable rate of decode-and-forward-based full-duplex MIMO relay systems, and propose a transmission scheme based on maximization of the lower bound. The maximization requires us to (numerically) solve a nonconvex optimization problem, for which we detail a novel approach based on bisection search and gradient projection. To gain insights into system design tradeoffs, we also derive an analytic approximation to the achievable rate and numerically demonstrate its accuracy. We then study the behavior of the achievable rate as a function of signal-to-noise ratio, interference-to-noise ratio, transmitter/receiver dynamic range, number of antennas, and training length, using optimized half-duplex signaling as a baseline.

This paper focuses on the downlink of a cellular system and studies opportunistic multiuser scheduling under imperfect channel state information, by exploiting the memory inherent in the channel. The channel between the base station and each user is modeled by a two-state Markov chain and the scheduled user sends back an ARQ feedback that arrives at the scheduler with a random delay, i.i.d. across users and time. The scheduler indirectly estimates the channel via accumulated delayed-ARQ feedback and uses this information to make scheduling decisions. The throughput maximization problem is formulated as a partially observable Markov decision process (POMDP). For the case of two users in the system, it is shown that a greedy policy is sum throughput optimal for any distribution on the ARQ feedback delay. For the case of more than two users, the greedy policy is suboptimal and numerical studies demonstrate that it has near optimal performance. Also, the greedy policy can be implemented by a simple algorithm that does not require the statistics of the underlying Markov channel or the ARQ feedback delay, thus making it robust against errors in system parameter estimation. Establishing an equivalence between the two-user system and a genie-aided system, a simple closed form expression for the sum capacity of the downlink is obtained. Further, inner and outer bounds on the capacity region of the downlink are obtained.

The problem of reliable communication over unknown frequency-selective block-fading channels with sparse impulse responses is considered. In particular, discrete-time impulse responses with block-fading interval N, length L; 0.1

This paper considers the problem of reliable communication over discrete-time channels whose impulse responses have lengthLand exactlyS≤Lnon-zero coefficients, and whose support and coefficients remain fixed over blocks ofN>;Lchannel uses but change independently from block to block. Here, it is assumed that the channel's support and coefficient realizations are both unknown, although their statistics are known. Assuming Gaussian non-zero-coefficients and noise, and focusing on the high-SNR regime, it is first shown that the ergodic noncoherent channel capacity has pre-log factor 1-(S)/(N) for any L. It is then shown that, to communicate with arbitrarily small error probability at rates in accordance with the capacity pre-log factor, it suffices to use pilot-aided orthogonal frequency-division multiplexing (OFDM) withSpilots per fading block, in conjunction with an appropriate noncoherent decoder. Since the achievability result is proven using a noncoherent decoder whose complexity grows exponentially in the number of fading blocksK, a simpler decoder, based onS+1 pilots, is also proposed. Its ε-achievable rate is shown to have pre-log factor equal to 1-(S+1)/(N) with the previously considered channel, while its achievable rate is shown to have pre-log factor 1-(S+1)/(N) when the support of the block-fading channel remains fixed over time.

In this paper, we consider noncoherent single-antenna communication over doubly selective block-fading channels with discrete block-fading intervalN. In our noncoherent setup, neither the transmitter nor the receiver know the channel fading coefficients, though both know the channel statistics. In particular, we consider discrete-time channels whose impulse-response trajectories obey a complex-exponential basis expansion model with uncorrelated coefficients, and we show that such a model holds in the limitN¿ ¿ for pulse-shaped transmission/reception over certain wide-sense stationary uncorrelated scattering channels. First, we show that, when the inputs are chosen from continuous distributions, the channel's multiplexing gain (i.e., capacity pre-log factor) equalsmax(0, 1 - NdelayNDopp/N) , for discrete delay spreadNdelayand discrete Doppler spreadNDopp. Next, for the case of strictly doubly selective fading (i.e.,NDopp> 1 andNdelay> 1), we establish that, for cyclic-prefixed affine pilot-aided transmission (PAT) schemes designed to minimize the mean-squared error (MSE) attained by pilot-aided minimum-MSE channel estimation, the pre-log factor of the achievable rate is less than the channel's multiplexing gain. We then provide guidelines for the design of PAT schemes whose achievable-rate pre-log factor equals the channel's multiplexing gain and construct an example.

We consider adapting the transmission rate to maximize the goodput, i.e., the amount of data transmitted without error, over a continuous Markov flat-fading wireless channel. In particular, we consider schemes in which transmitter channel state is inferred from degraded causal error-rate feedback, such as packet-level ACK/NAKs in an automatic repeat request (ARQ) system. In such schemes, the choice of transmission rate affects not only the subsequent goodput but also the subsequent feedback, implying that the optimal rate schedule is given by a partially observable Markov decision process (POMDP). Because solution of the POMDP is computationally impractical, we consider simple suboptimal greedy rate assignment and show that the optimal scheme would itself be greedy if the error-rate feedback was non-degraded. Furthermore, we show that greedy rate assignment using non-degraded feedback yields a total goodput that upper bounds that of optimal rate assignment using degraded feedback. We then detail the implementation of the greedy scheme and propose a reduced-complexity greedy scheme that adapts the transmission rate only once per block of packets. We also investigate the performance of the schemes numerically, and show that the proposed greedy scheme achieves steady-state goodputs that are reasonably close to the upper bound on goodput calculated using non-degraded feedback. A similar improvement is obtained in steady-state goodput, drop rate, and average buffer occupancy in the presence of data buffers. We also investigate an upper bound on the performance of optimal rate assignment for a discrete approximation of the channel and show that such quantization leads to a significant loss in achievable goodput.

In this paper we propose a novel method for communication over underwater acoustic channels that exhibit simultaneously large delay spread and Doppler spread, such as those found in the surf zone. In particular, we propose a coded pulse-shaped multicarrier scheme that converts the doubly dispersive channel into an inter-carrier interference (ICI) channel with small ICI spread. The resulting ICI is mitigated using a soft noncoherent equalizer that leverages sparsity in the delay-power profile to generate near-optimal bit estimates with low complexity. The noncoherent equalizer uses a delay-power-profile estimate (rather than a channel estimate) which is obtained from pilots. Numerical simulations with surf-zone-like channels demonstrate performance close to genie-aided bounds.

In this correspondence, the performance of the automatic repeat request-dynamic decode and forward (ARQ-DDF) cooperation protocol is analyzed in two distinct scenarios. The first scenario is the multiple access relay channel where a single relay is dedicated to simultaneously help two multiple access users. For this setup, it is shown that the ARQ-DDF protocol achieves the channel's optimal diversity multiplexing tradeoff (DMT). The second scenario is the cooperative vector multiple access channel where two users cooperate in delivering their messages to a destination equipped with two receiving antennas. For this setup, a new variant of the ARQ-DDF protocol is developed where the two users are purposefully instructed not to cooperate in the first round of transmission. Lower and upper bounds on the achievable DMT are then derived. These bounds are shown to converge to the optimal tradeoff as the number of transmission rounds increases.

Compared to conventional time-domain equalization, frequency-domain equalization (FDE) presents a computationally efficient alternative for the reception of single carrier (SC) transmissions. In this paper, we consider iterative FDE (IFDE) with explicit frequency-domain channel estimation (FDCE) for non-cyclic-prefixed SC systems. First, an improved IFDE algorithm is presented based on soft iterative interference cancellation. Second, a new adaptive FDCE (AFDCE) algorithm based on per-tone Kalman filtering is proposed to track and predict the frequency-domain channel coefficients. The AFDCE algorithm employs across-tone noise reduction, exploits temporal correlation between successive blocks, and adaptively updates the auto-regressive model coefficients, bypassing the need for prior knowledge of channel statistics. Finally, block-overlapping is used to facilitate the joint operation of IFDE and AFDCE. Simulation results show that, compared to related IFDE and adaptive channel estimation schemes, the proposed schemes offer lower mean-square error (MSE) in channel prediction, lower bit error rate (BER) after decoding, and robustness to non-stationary channels.

Channel state information (CSI) is important for achieving large rates in MIMO channels. However, in time-varying MIMO channels, there is a tradeoff between the time/energy spent acquiring channel state information (CSI) and the time/energy remaining for data transmission. This tradeoff is accentuated in the MIMO multiple access channel (MAC), since the number of channel vectors to be estimated increases with the number of users. Furthermore, the problem of acquiring CSI is tightly coupled with the problem of exploiting CSI through multiuser scheduling. This paper considers a block-fading MAC with coherence timeT, n uncoordinated users-each with one transmit antenna and the same average power constraint, and a base station withMreceive antennas and no a priori CSI. For this scenario, a training-based communication scheme is proposed and the training and multiuser-scheduling aspects of the scheme are jointly optimized. In the high-SNR regime, the sum capacity of the non-coherent SIMO MAC is characterized and used to establish the SNR-scaling-law optimality of the proposed scheme. In the low-SNR regime, the sum-rate of the proposed scheme is found to decay linearly with vanishing SNR when flash signaling is incorporated. Furthermore, this linear decay is shown to be order-optimal through comparison to the low-SNR sum capacity of the non-coherent SIMO MAC. A by product of these SNR-asymptotic analyses is the observation that non-trivial scheduling (i.e., scheduling a strict subset of trained users) is advantageous at low SNR, but not at high SNR. The sum-rate and per-user throughput are also explored in the large-nand large-Mregimes. Non-coherent capacity, training, multiple access channel, multiuser scheduling, opportunistic scheduling.

We consider the effect of mobility on a wideband direct sequence spread spectrum (DSSS) communication system, and study a scale-lag Rake receiver capable of leveraging the diversity that results from mobility. A wideband signal has a large bandwidth-to-center frequency ratio, such that the typical narrowband Doppler spread assumptions do not apply to mobile channels. Instead, we assume a more general temporal scaling phenomenon, i.e., a dilation of the transmitted signal's time support. Based on a uniform ring of scatterers model, we determine that the wideband scattering function, which quantifies the average scale spreading, has a "bathtub-shaped" scale profile. We compare the performances of a scale-lag Rake and a frequency-lag Rake, each capable of leveraging the diversity that results from mobility. Such analysis applies, for example, to ultra-wideband (UWB) radio frequency channels and underwater wideband acoustic channels.

We consider pilot-aided transmission (PAT) for a general class of systems encompassing linear modulation and a linear time-varying channel. For these systems, and given a pilot energy constraint, we derive a tight lower bound on the mean squared error (MSE) of pilot-aided channel estimates as well as necessary and sufficient conditions on PAT to attain this bound. We then apply these results to the design of single-antenna PAT for doubly selective channels and arrive at novel MSE-optimal PAT schemes. In this application, we assume a block-based cyclic-prefix PAT and a basis expansion model for the channel.

We consider the effect of mobility on a wideband direct sequence spread spectrum (DSSS) communication system, and study a scale-lag Rake receiver capable of leveraging the diversity that results from mobility. A wideband signal has a large bandwidth-to-center frequency ratio, such that the typical narrowband Doppler spread assumptions do not apply to mobile channels. Instead, we assume a more general temporal scaling phenomenon, i.e., a dilation of the transmitted signal's time support. Such analysis applies, for example, to ultra-wideband (UWB) radio frequency channels and underwater wideband acoustic channels.

We propose novel cooperative transmission protocols for delay-limited coherent fading channels consisting of N (half-duplex and single-antenna) partners and one cell site. In our work, we differentiate between the relay, cooperative broadcast (down-link), and cooperative multiple-access (CMA) (up-link) channels. The proposed protocols are evaluated using Zheng-Tse diversity-multiplexing tradeoff. For the relay channel, we investigate two classes of cooperation schemes; namely, amplify and forward (AF) protocols and decode and forward (DF) protocols. For the first class, we establish an upper bound on the achievable diversity-multiplexing tradeoff with a single relay. We then construct a new AF protocol that achieves this upper bound. The proposed algorithm is then extended to the general case with (N-1) relays where it is shown to outperform the space-time coded protocol of Laneman and Wornell without requiring decoding/encoding at the relays. For the class of DF protocols, we develop a dynamic decode and forward (DDF) protocol that achieves the optimal tradeoff for multiplexing gains 0lesrles1/N. Furthermore, with a single relay, the DDF protocol is shown to dominate the class of AF protocols for all multiplexing gains. The superiority of the DDF protocol is shown to be more significant in the cooperative broadcast channel. The situation is reversed in the CMA channel where we propose a new AF protocol that achieves the optimal tradeoff for all multiplexing gains. A distinguishing feature of the proposed protocols in the three scenarios is that they do not rely on orthogonal subspaces, allowing for a more efficient use of resources. In fact, using our results one can argue that the suboptimality of previously proposed protocols stems from their use of orthogonal subspaces rather than the half-duplex constraint.

We consider the design of max-SINR pulse-shaped (PS) frequency domain modulation (FDM), where the signal to interference-plus-noise ratio (SINR) is defined in accordance with inter-symbol and intercarrier interference (ISI/ICI) shaping rather than complete ISI/ICI suppression. Because the transmitter is assumed to know the channel scattering function but not the channel realization, the resulting max-SINR pulses are non-(bi)orthogonal. For this case, numerical results suggest that max-SINR systems designed for ISI/ICI-shaping achieve higher outage capacity than those designed for ISI/ICI-suppression. An outage capacity analysis is also used to obtain rough design guidelines for max-SINR non-(bi)orthogonal PS-FDM, since the design paradigm differs from that of (bi)orthogonal PS-FDM.

In this paper, we propose novel cooperative transmission protocols for delay limited coherent fading channels consisting of N (half-duplex and single-antenna) partners and one cell site. In our work, we differentiate between the cooperative relay, broadcast, and multiple-access channels. The proposed protocols are evaluated using the Zheng-Tse diversity-multiplexing tradeoff. For the relay channel, we investigate two classes of cooperation schemes; namely, amplify and forward (AF) and decode and forward (DF). For the first class, we propose a new AF protocol and show it to outperform the space-time coded protocol of Laneman and Wornell without requiring decoding/encoding at the relays. For the class of DF protocols, we develop a dynamic decode and forward (DDF) protocol that achieves the optimal tradeoff for multiplexing gains 0 /spl les/ r /spl les/ 1/N. Furthermore, with a single relay, the DDF protocol is shown to dominate the class of AF protocols for all multiplexing gains. The superiority of the DDF protocol is shown to be more significant in the cooperative broadcast channel. The situation is reversed in the cooperative multiple-access channel where we propose a new AF protocol that achieves the optimal tradeoff for all multiplexing gains. A distinguishing feature of the proposed protocols in the three scenarios is that they do not rely on orthogonal subspaces, allowing for a more efficient use of resources.

The least mean squared (LMS) adaptive filtering algorithm may experience uncontrolled parameter drift when its input signal is not persistently exciting, leading to serious consequences when implemented with finite word-length. Though so-called "tap-leakage" modifications of LMS have been proposed to mitigate this drift, they inevitably introduce parameter bias which degrades mean-squared error performance. In this letter, we propose a novel algorithm which leaks only in the unexcited modes, thus introducing insignificant bias, while still retaining the low computational complexity of LMS.

Decision-feedback differential detection (DFDD) of differential phase-shift keying (DPSK) and differential unitary space-time modulation (DUST) in Rayleigh-fading channels exhibits significant performance improvement over standard single-symbol maximum-likelihood detection. However, knowledge of channel fading correlation and signal-to-noise ratio (SNR) is required at the receiver to compute the feedback coefficients used in DFDD. In this letter, we investigate the robustness of the DFDD to imperfect knowledge of the feedback coefficients by modeling the mismatch between estimated feedback coefficients and ideal coefficients in terms of mismatch between the estimated values of fading correlation and SNR and the true values. Under the assumption of a block-fading channel when nondiagonal DUST constellations are used and a continuous fading channel otherwise, we derive exact and Chernoff bound expressions for pair-wise word-error probability and then use them to approximate the bit-error rate (BER), finding close agreement with simulation results. The relationships between BER performance and various system parameters, e.g., DFDD length and Doppler mismatch, are also explored. Furthermore, the existence of an error floor in the BER-vs-SNR curve is investigated for the infinite-length DFDD. For the special case of Jakes' fading model, it is shown that the error floor can be removed completely even when the Doppler spread is over-estimated.

Orthogonal frequency division multiplexing (OFDM) systems may experience significant intercarrier interference (ICI) when used in time- and frequency-selective, or doubly-selective, channels. In such cases, the classical symbol estimation schemes, e.g., minimum mean-squared error (MMSE) and zero-forcing (ZF) estimation, require matrix inversion that is prohibitively complex for large symbol lengths. An analysis of the ICI generation mechanism leads us to propose a novel two-stage equalizer whose complexity (apart from the FFT) is linear in the OFDM symbol length. The first stage applies optimal linear preprocessing to restrict ICI support and the second stage uses iterative MMSE estimation to estimate finite-alphabet frequency-domain symbols. Simulation results indicate that our equalizer has significant performance and complexity advantages over the classical linear MMSE estimator in doubly-selective channels.

Decision-feedback differential detection (DFDD) of differential phase-shift keying (DPSK) and differential unitary space-time modulation (DUST) in Rayleigh fading channels exhibits significant performance improvement over standard single-symbol ML detector, but it requires the knowledge of fading correlation and SNR at the receiver. In this paper, we investigate the robustness of the DFDD to imperfect knowledge of fading correlation and SNR. We derive the exact and Chernoff bound expressions for pair-wise word-error probability (PWEP), and then use them to approximate the BER, finding close agreement with simulation results. The relationships between performance and various system parameters, e.g., DFDD length and Doppler mismatch are also explored.

Connexions is a new approach to authoring, teaching, and learning that aims to fully exploit modern information technology. Available free of charge to anyone under open-content and open-source licenses, Connexions offers custom-tailored, current course material, is adaptable to a wide range of learning styles, and encourages students to explore the links among courses and disciplines. In contrast to the traditional process of textbook writing and publishing, Connexions fosters world-wide, cross-institution communities of authors, instructors, and students, who collaboratively and dynamically fashion “modules” from which courses are constructed. We believe the ideas and philosophy embodied by Connexions have the potential to change the very nature of textbook writing and publishing, producing a dynamic, interconnected educational environment that is pedagogically sound, both time and cost efficient, and fun. This paper overviews the philosophy and technology behind Connexions and describes a nascent community developing material for DSP education.

This correspondence shows the equivalence of three previously proposed reduced-rank detection schemes for direct-sequence code-division multiple-access (DS-CDMA) communication systems. The auxiliary vector filtering (AVF) algorithm is simplified through a key observation on the construction of the auxiliary vectors. After simplification, it is shown that the AVF algorithm is equivalent to the multistage Wiener filtering (MWF) algorithm of Honig and Goldstein (2002). Furthermore, these schemes can be shown to be equivalent to the multistage linear receiver scheme based on the Cayley-Hamilton (CH) theorem when the minimum mean-square error (MMSE) criterion is applied to the reduced dimensional space of the received signal.

The constant modulus (CM) criterion has become popular in the design of blind linear estimators of sub-Gaussian independent and identically distributed (i.i.d.) processes transmitted through unknown linear channels in the presence of unknown additive interference. In this paper, we present an upper bound for the conditionally unbiased mean-squared error (UMSE) of CM-minimizing estimators that depends only on the source kurtoses and the UMSE of Wiener estimators. Further analysis reveals that the extra UMSE of CM estimators can be upper-bounded by approximately the square of the Wiener (i.e., minimum) UMSE. Since our results hold for vector-valued finite-impulse response/infinite-impulse response (FIR/IIR) linear channels, vector-valued FIR/IIR estimators with a possibly constrained number of adjustable parameters, and multiple interferers with arbitrary distribution, they confirm the longstanding conjecture regarding the general mean-square error (MSE) robustness of CM estimators.

The constant modulus (CM) criterion has become popular in the design of blind linear estimators of sub-Gaussian i.i.d. processes transmitted through unknown linear channels in the presence of unknown additive interference. The existence of CM local minima, however, makes it difficult for CM-minimizing schemes to generate estimates of the desired source (as opposed to an interferer) in multiuser environments. In this paper, we present three sufficient conditions under which gradient descent (GD) minimisation of CM cost will locally converge to an estimator of the desired source at a particular delay. The sufficient conditions are expressed in terms of statistical properties of the initial estimates, specifically, CM cost, kurtosis, and signal to interference-plus-noise ratio (SINR).

Adaptive blind equalization has gained widespread use in communication systems that operate without training signals. In particular, the constant modulus algorithm (CMA) has become a favorite of practitioners due to its LMS-like complexity and desirable robustness properties. The desire for further reduction in computational complexity has motivated signed-error versions of CMA, which have been found to lack the robustness properties of the CMA. This paper presents a simple modification of signed-error CMA, based on the judicious use of dither, that results in an algorithm with robustness properties closely resembling those of CMA. An approximation to the steady-state mean-squared error performance of the new algorithm is derived for comparison to that of CMA.

This paper provides a tutorial introduction to the constant modulus (CM) criterion for blind fractionally spaced equalizer (FSE) design via a (stochastic) gradient descent algorithm such as the constant modulus algorithm (CMA). The topical decisions utilized in this tutorial can be used to help catalog the emerging literature on the CM criterion and on the behavior of (stochastic) gradient descent algorithms used to minimize it.