In this paper, we study the impacts of transmit antenna selection on secrecy performance massive MIMO systems. We consider a wiretap setting in which fixed number antennas are selected and then confidential messages transmitted over them to multi-antenna legitimate receiver while being overheard by eavesdropper. For setup, derive an accurate approximation instantaneous rate. Using approximation, it is shown that some settings under growth active enhances system up optimal degrades when...

In intelligent reflecting surface (IRS)-aided multiple-input multiple-output (MIMO) systems, the IRS can be utilized to suppress information leakage towards malicious terminals. This lead significant secrecy gains. work exploits these gains via a tractable <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">joint</i> design of downlink beamformers and phase-shifts. this respect, we consider generic IRS-aided MIMO wiretap setting invoke fractional...

In federated learning (FL), heterogeneity among the local dataset distributions of clients can result in unsatisfactory performance for some, leading to an unfair model. To address this challenge, we propose over-the-air fair algorithm (OTA-FFL), which leverages computation train FL models. By formulating as a multi-objective minimization problem, introduce modified Chebyshev approach compute adaptive weighting coefficients gradient aggregation each communication round. enable efficient over...

This work invokes the notion of $f$-divergence to introduce a novel upper bound on Bayes error rate general classification task. We show that proposed can be computed by sampling from output parameterized model. Using this practical interpretation, we optimal learning threshold (BOLT) loss whose minimization enforces model achieve rate. validate for image and text tasks, considering MNIST, Fashion-MNIST, CIFAR-10, IMDb datasets. Numerical experiments demonstrate models trained with BOLT...

Pinching antennas have been recently proposed as a promising flexible-antenna technology, which can be implemented by attaching low-cost pinching elements to dielectric waveguides. This work explores the potential of employing antenna systems (PASs) for downlink transmission in multiuser MIMO setting. We consider problem hybrid beamforming, where digital precoder at access point and activated locations are jointly optimized maximize achievable weighted sum-rate. Invoking fractional...

Error accumulation is effective for gradient sparsification in distributed settings: initially-unselected entries are eventually selected as their accumulated error exceeds a certain level. The essentially behaves scaling of the learning rate entries. Although this property prevents slow-down lateral movements descent, it can deteriorate convergence some settings. This work proposes novel scheme that controls accumulation. development follows two major steps: first, formulated an inverse...

This paper proposes the class of generalized least-square-error (GLSE) precoders for multiuser massive multiple-input multiple-output (MIMO) systems. For a generic transmit constellation, GLSE minimize interference at user terminals assuring that some given constraints on signals are satisfied. The general form these enables us to impose multiple restrictions signal, such as limited peak power and restricted number active antennas. performance is analyzed in large-system limit. It shown...

This work studies secure transmission in intelligent reflecting surface (IRS)-assisted MIMO systems when an active eavesdropper is available the network. We consider a scenario which performs pilot attack to contaminate channel estimation at base station. Invoking method of regularized zero forcing, we develop algorithm that designs beamforming vectors, as well phase-shifts IRS, such attacker blinded. Our numerical investigations confirm proposed can suppress effectively, long legitimate and...

This paper develops a class of low-complexity device scheduling algorithms for over-the-air federated learning via the method matching pursuit. The proposed scheme tracks closely <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">close-to-optimal</i> performance achieved by difference-of-convex programming, and outperforms significantly well-known benchmark based on convex relaxation. Compared to state-of-the-art, imposes drastically lower...

This paper studies the large-system performance of Least Square Error (LSE) precoders which minimize input-output distortion over an arbitrary support subject to a general penalty function. The asymptotics are determined via replica method in form encloses Replica Symmetric (RS) and Symmetry Breaking (RSB) ansätze. As result, "marginal decoupling property" LSE for b-steps RSB is derived. generality studied setup enables us address special cases number active transmit antennas constrained....

The large-system performance of maximum-a-poste-rior estimation is studied considering a general distortion function when the observation vector received through linear system with additive white Gaussian noise. analysis considers matrix to be chosen from large class rotationally invariant random matrices. We take statistical mechanical approach by introducing spin glass corresponding estimator, and employing replica method for analysis. In contrast earlier based studies, our evaluates...

In massive MIMO wiretap settings, the base station can significantly suppress eavesdroppers by narrow beamforming toward legitimate terminals. Numerical investigations show that this approach, secrecy is obtained at no significant cost. We call property of systems "secrecy for free" and it not only holds when all transmit antennas are employed, but also a single antenna set active. Using linear precoding, information leakage to be sufficiently diminished, total number available grows large,...

It is widely believed that large IRS-aided MIMO settings maintain the fundamental features of massive systems. This work gives a rigorous proof confirms this belief. We show using passive IRS, end-to-end channel between transmitter and receiver always hardens, even if IRS elements are strongly correlated. For fading direct reflection links receiver, our derivations demonstrate for number reflecting on capacity accurately approximated by real-valued Gaussian random variable whose variance...

In this paper, we investigate the problem of empirical coordination in a triangular multiterminal network. A network consists three terminals where two observe external i.i.d correlated sequences. The third terminal wishes to generate sequence with desired joint distribution. For problem, derive inner and outer bounds on capacity region. It is shown that region degraded source cascade can be directly obtained from our bounds. cipher system, establish key distribution over reliable terminal,...

This paper proposes nonlinear least square error (LSE) precoders for multiuser MIMO broadcast channels. The LSE are designed such that the discrete output signals from a predefined set. set allows us to model several signal constraints as peak power constraint, constant envelope, and constellations. We study large-system performance of these via replica method statistical physics, derive closed-form expression asymptotic distortion. Our results demonstrate an precoder with peak-to-average...

Consider a fading Gaussian MIMO channel with N <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">t</sub> transmit and xmlns:xlink="http://www.w3.org/1999/xlink">r</sub> receive antennas. The transmitter selects L antennas corresponding to the strongest channels. For this setup, we study distribution of input-output mutual information when grows large. We show that, for any , is accurately approximated by whose mean large variance converges zero....

The large-system decoupling property of a MAP estimator is studied when it estimates the i.i.d. vector $\boldsymbol{x}$ from observation $\boldsymbol{y}=\mathbf{A}\boldsymbol{x}+\boldsymbol{z}$ with $\mathbf{A}$ being chosen wide range matrix ensembles, and noise $\boldsymbol{z}$ Gaussian. Using replica method, we show that marginal joint distribution any two corresponding input output symbols converges to deterministic which describes input-output single user system followed by estimator....

In this work, we propose a new single-RF MIMO architecture which enjoys high scalability and energy-efficiency. The transmitter in proposal consists of single RF illuminator radiating towards reflecting surface. Each element on the surface re-transmits its received signal after applying phase-shift, such that desired beamforming pattern is obtained. For architecture, problem interpreted as linear regression solution derived via method least-squares. Using formulation, fast iterative...

The large-system performance of MAP estimation is studied considering a general distortion function when the observation vector received through linear system with additive white Gaussian noise. analysis considers matrix to be chosen from large class rotationally invariant random matrices. We take statistical mechanical approach by introducing spin glass corresponding estimator, and employing replica method for analysis. In contrast earlier based studies, our evaluates ansatz determines...

Due to the large power consumption in RF-circuitry of massive MIMO systems, practically relevant performance measures such as energy efficiency or bandwidth are neither necessarily monotonous functions total transmit nor number active antennas. Optimal antenna selection is however computationally infeasible these systems. In this paper, we propose an iterative algorithm optimize and subset selected antennas subject non-monotonous multiuser settings. Numerical results given for demonstrate...