In Federated Learning (FL) with over-the-air aggregation, the quality of the signal received at the server critically depends on the receive scaling factors. While a larger scaling factor can reduce the effective noise power and improve training performance, it also compromises the privacy of devices by reducing uncertainty. In this work, we aim to adaptively design the receive scaling factors across training rounds to balance the trade-off between training convergence and privacy in an FL system under dynamic channel conditions. We formulate a stochastic optimization problem that minimizes the overall Rényi differential privacy (RDP) leakage over the entire training process, subject to a long-term constraint that ensures convergence of the global loss function. Our problem depends on unknown future information, and we observe that standard Lyapunov optimization is not applicable. Thus, we develop a new online algorithm, termed AdaScale, based on a sequence of novel per-round problems that can be solved efficiently. We further derive upper bounds on the dynamic regret and constraint violation of AdaSacle, establishing that it achieves diminishing dynamic regret in terms of time-averaged RDP leakage while ensuring convergence of FL training to a stationary point. Numerical experiments on canonical classification tasks show that our approach effectively reduces RDP and DP leakages compared with state-of-the-art benchmarks without compromising learning performance.

This paper studies power-efficient uplink transmission design for federated learning (FL) that employs over-the-air analog aggregation and multi-antenna beamforming at the server. We jointly optimize device transmit weights and receive beamforming at each FL communication round to minimize the total device transmit power while ensuring convergence in FL training. Through our convergence analysis, we establish sufficient conditions on the aggregation error to guarantee FL training convergence. Utilizing these conditions, we reformulate the power minimization problem into a unique bi-convex structure that contains a transmit beamforming optimization subproblem and a receive beamforming feasibility subproblem. Despite this unconventional structure, we propose a novel alternating optimization (AO) approach that guarantees monotonic decrease of the objective value, to allow convergence to a partial optimum. We further consider imperfect channel state information (CSI), which requires accounting for the channel estimation errors in the power minimization problem and FL convergence analysis. We propose a CSI-error-aware joint beamforming algorithm, which can substantially outperform one that does not account for channel estimation errors. Simulation with canonical classification datasets demonstrates that our proposed methods achieve significant power reduction compared to existing benchmarks across a wide range of parameter settings, while attaining the same target accuracy under the same convergence rate.

Federated learning (FL) over wireless networks using analog transmission can efficiently utilize the communication resource but is susceptible to errors caused by noisy wireless links. In this paper, assuming a multi-antenna base station, we jointly design downlink-uplink beamforming to maximize FL training convergence over time-varying wireless channels. We derive the round-trip model updating equation and use it to analyze the FL training convergence to capture the effects of downlink and uplink beamforming and the local model training on the global model update. Aiming to maximize the FL training convergence rate, we propose a low-complexity joint downlink-uplink beamforming (JDUBF) algorithm, which adopts a greedy approach to decompose the multi-round joint optimization and convert it into per-round online joint optimization problems. The per-round problem is further decomposed into three subproblems over a block coordinate descent framework, where we show that each subproblem can be efficiently solved by projected gradient descent with fast closed-form updates. An efficient initialization method that leads to a closed-form initial point is also proposed to accelerate the convergence of JDUBF. Simulation demonstrates that JDUBF substantially outperforms the conventional separate-link beamforming design.

In this paper, we derive the Cramer-Rao lower bound (CRLB) for a newly developed approach for direction of arrival (DoA) estimation in hybrid analog/digital (HAD) receivers under snapshot constraints. In such cases, the inherent structure of the received signals can be exploited for reliable DoA estimation rather than using statistical averaging techniques. One approach to exploit this structure is the matrix pencil method (MPM). Unfortunately, existing HAD receivers tangle the signals at the output of the HAD receiver, hindering the direct use of the MPM. To address this difficulty, an approach developed in [1] enables the MPM to expose the structure of the output signal of the analog combiner by leveraging periodic, potentially unknown signals to disentangle the output of the HAD receiver. We derive the CRLB for this approach and show that it yields output signals resembling those of a fully-digital receiver, albeit with a snapshot penalty. Numerical simulations show that the developed approach achieves performance within a small gap of the corresponding CRLB and outperforms existing counterparts.

This work addresses the trade-off between convergence and the overall delay in heterogeneous distributed learning systems, where the devices encounter diverse and dynamic communication conditions. We propose to apply adaptive sparsification across the devices and over iterations, formulating an optimization problem to minimize the overall delay while ensuring a specified level of convergence. The resultant stochastic optimization problem cannot be handled by conventional Lyapunov optimization techniques due to the dependency of the per-iteration objective function on the previous iterations. To overcome this challenge, we propose AdaSparse, an online algorithm with a novel per-slot problem that can be solved optimally by searching over a finite discrete space. We further introduce a low-complexity approximation of AdaSparse, termed LC-AdaSparse, which features linear computational complexity and diminishing approximation error. We show that AdaSparse offers strong performance guarantees, simultaneously achieving sub-linear dynamic regret in terms of delay and the optimal rate in terms of convergence. Numerical experiments on classification tasks using standard datasets and various models demonstrate that our approach effectively reduces the communication delay compared with existing benchmarks, to achieve the same levels of learning accuracy.

Federated learning (FL) with over-the-air computation efficiently utilizes the communication resources, but it can still experience significant latency when each device transmits a large number of model parameters to the server. This paper proposes the Segmented Over-The-Air (SegOTA) method for FL, which reduces latency by partitioning devices into groups and letting each group transmit only one segment of the model parameters in each communication round. Considering a multiantenna server, we model the SegOTA transmission and reception process to establish an upper bound on the expected model learning optimality gap. We minimize this upper bound, by formulating the per-round online optimization of device grouping and joint transmit-receive beamforming, for which we derive efficient closed-form solutions. Simulation results show that our proposed SegOTA substantially outperforms the conventional full-model OTA approach and other common alternatives.

This paper studies the design of wireless federated learning (FL) for simultaneously training multiple machine learning models. We consider round robin device-model assignment and downlink beamforming for concurrent multiple model updates. After formulating the joint downlink-uplink transmission process, we derive the per-model global update expression over communication rounds, capturing the effect of beamforming and noisy reception. To maximize the multi-model training convergence rate, we derive an upper bound on the optimality gap of the global model update and use it to formulate a multi-group multicast beamforming problem. We show that this problem can be converted to minimizing the sum of inverse received signal-to-interference-plus-noise ratios, which can be solved efficiently by projected gradient descent. Simulation shows that our proposed multi-model FL solution outperforms other alternatives, including conventional single-model sequential training and multi-model zero-forcing beamforming.

We consider wireless network virtualization (WNV) in an uplink multiple-input multiple-output system, where multiple service providers (SPs) operate in virtually isolated networks managed by an infrastructure provider (InP) that owns the communication equipment. Service isolation is achieved at the physical layer by exploiting a large number of antennas at the base stations. We formulate this WNV as a non-convex optimization problem for the InP, jointly considering the uplink receive beamforming at the BS and the transmit power of the SPs' subscribing user devices. We decompose the problem into two subproblems and derive closed-form solutions to both. We then adopt an alternating optimization approach to combine the closed-form solutions to solve the original problem. Our simulation results show that the proposed method provides strong service isolation among the SPs while retaining efficiency similar to or better than centralized beamforming without virtualization, and it substantially outperforms traditional WNV with strict resource separation.

Placing selected content at the edge of the network close to the users, known as caching, is an important technique to improve the efficiency of content delivery in wireless networks. In this paper, we consider caching in a cloud radio access network (C-RAN) in which the primary fronthaul link operates in the mmWave range and may switch to microwave frequencies in the case of blockage. We aim to minimize the average long-term network cost by optimizing dynamic fetching and caching decisions. Importantly, we consider the realistic case of user request distributions and blockage rates being a priori unknown and not necessarily stationary. We introduce change point detection (CPD) to detect significant changes in the environment; we couple this step with reinforcement learning (RL): our key contribution, the proposed change point detection assisted reinforcement learning (CPRL) algorithm learns the environment and (re-)optimizes the caching policy to solve the associated Markov decision process (MDP) problem. Essentially, CPD allows our learning algorithm to adapt its caching strategy to the new environment which shows faster convergence. The numerical results show that our proposed approach improves the efficiency of caching in wireless networks, making it more adaptable to changing request patterns over time.

We propose a low-complexity downlink channel covariance matrix estimation for massive multiple-input multiple-output systems in which the base station (BS) is equipped with a uniform rectangular antenna array (URA). This scheme can be expressed in the form of an affine transformation which depends only on the uplink and downlink carrier frequencies, and the BS array configurations. An upper bound on the estimation error is derived, which shows that the accuracy of the proposed scheme increases with the number of URA antennas and the compactness and differentiability class of the periodic extension of a non-linearly transformed version of the angular power spread. The performance superiority of the proposed scheme over its existing counterparts is confirmed through simulations.

Combining over-the-air uplink transmission and multi-antenna beamforming can improve the efficiency of federated learning (FL). However, to mitigate the significant aggregation error due to communication noise and signal distortion, pre-processing of device signals and post-processing at the server are required. In this paper, we study the optimization of receiver beamforming and device transmit weights in over-the-air FL, to minimize the total transmit power in each communication round while guaranteeing the convergence of FL. We establish sufficient convergence conditions based on the analysis of gradient descent with error and formulate a power minimization problem. An alternating optimization approach is then employed to decompose the problem into tractable subproblems, and efficient solutions are developed for these subproblems. Our proposed method is evaluated through simulation on standard image classification tasks, demonstrating its effectiveness in achieving substantial reductions in transmit power compared with existing alternatives.

We study joint downlink-uplink beamforming design for wireless federated learning (FL) with a multi-antenna base station. Considering analog transmission over noisy channels and uplink over-the-air aggregation, we derive the global model update expression over communication rounds. We then obtain an upper bound on the expected global loss function, capturing the downlink and uplink beamforming and receiver noise effect. We propose a low-complexity joint beamforming algorithm to minimize this upper bound, which employs alternating optimization to breakdown the problem into three subproblems, each solved via closed-form gradient updates. Simulation under practical wireless system setup shows that our proposed joint beamforming design solution substantially outperforms the conventional separate-link design approach and nearly attains the performance of ideal FL with error-free communication links.

An efficient algorithm for identifying space-time block-coded orthogonal frequency division multiplexing (STBC-OFDM) signals is introduced in this paper. The inherited signal redundancy is exploited for identification, with the second-order correlations between pairs of signals received from diverse antennas used as the identification feature. The decision on the STBC signal is made by employing the statistical properties of the feature estimate. The proposed algorithm does not require STBC or OFDM block synchronization, channel or noise power estimation, and knowledge of the signal constellation. The performance of the proposed algorithm is evaluated through extensive simulation experiments. The results show the superiority of the proposed algorithm over the previously reported algorithms, with a reduced observation time and at lower signal-to-noise ratio.

This paper proposes an efficient identification algorithm for spatial multiplexing (SM) and Alamouti (AL) coded orthogonal frequency-division multiplexing (OFDM) signals. The cross correlation between the received signals from different antennas is exploited to provide a discriminating feature to identify SM-OFDM and AL-OFDM signals. The proposed algorithm requires neither estimation of the channel coefficients and noise power, nor the modulation of the transmitted signal. Moreover, it does not need space-time block code or OFDM block synchronization. The effectiveness of the proposed algorithm is demonstrated through extensive simulation experiments in the presence of diverse transmission impairments, such as time and frequency offsets, Doppler frequency, and spatially correlated fading.

In this paper, an algorithm for the detection of the Alamouti-coded orthogonal frequency division multiplexing (AL-OFDM) signals is proposed. To the best of our knowledge, this is the first time in the literature when the detection of AL-OFDM signals used in recent WiMAX and LTE standards is investigated. The cross-correlation between the signals received with two antennas is studied as a detection feature, and its analytical closed-form expression obtained. These findings are further employed to develop the signal detection algorithm. The algorithm performance is investigated based on simulated standard signals. A good performance is achieved with a short sensing time and at low signal-to-noise ratios (SNRs). Additionally, the proposed algorithm requires neither information about the channel, modulation type, and noise power, nor timing synchronization.

This paper proposes a novel and efficient algorithm for space-time block code (STBC) classification, when a single antenna is employed at the receiver. The algorithm exploits the discriminating features provided by the discrete Fourier transform (DFT) of the fourth-order lag products (FOLPs) of the received signal. It does not require estimation of the channel, signal-to-noise ratio (SNR), and modulation of the transmitted signal. Computer simulations are conducted to evaluate the performance of the proposed algorithm. The results show the validity of the algorithm, its robustness to carrier frequency offset, and low sensitivity to timing offset.

Blind signal identification is an important topic of research for both commercial and military communications. A novel identification algorithm for spatial multiplexing (SM) and Alamouti space-time block code (AL-STBC) signals is proposed in this paper, when the receiver is equipped with a single antenna. The proposed algorithm exploits a discriminating feature provided by the discrete Fourier transform (DFT) of the fourth-order lag product (FOLP) of the received signal. The proposed algorithm requires neither estimation of the channel, noise power, nor modulation of the transmitted signal. Computer simulations are conducted to evaluate the algorithm performance; these show the validity of the proposed algorithm with low sensitivity to timing offset.

Blind signal classification, a major task of intelligent receivers, has important civilian and military applications. This problem becomes more challenging in multi-antenna scenarios due to the diverse transmission schemes that can be employed, e.g., spatial multiplexing (SM) and space-time block codes (STBCs). This paper presents a class of novel algorithms for blind classification of SM and Alamouti STBC (AL-STBC) transmissions. Unlike the prior art, we show that signal classification can be performed using a single receive antenna by taking advantage of the space-time redundancy. The first proposed algorithm relies on the fourth-order moment as a discriminating feature and employs the likelihood ratio test for achieving maximum average probability of correct classification. This requires knowledge of the channel coefficients, modulation type, and noise power. To avoid this drawback, three algorithms have been further developed. Their common idea is that the discrete Fourier transform of the fourth-order lag product exhibits peaks at certain frequencies for the AL-STBC signals, but not for the SM signals, and thus, provides the basis of a useful discriminating feature for signal classification. The effectiveness of these algorithms has been demonstrated in extensive simulation experiments, where a Nakagami-m fading channel and the presence of timing and frequency offsets are assumed.

This paper investigates the second-order cyclostationarity of block transmitted-single carrier linearly digitally modulated (BT-SCLD) signals, and its applications to signal classification and blind (non-data aided) parameter estimation. Analytical closed-form expressions are derived for the cyclic autocorrelation function (CAF), cyclic spectrum (CS), complementary CAF (CCAF), complementary CS (CCS), and corresponding cycle frequencies (CFs). Furthermore, the conditions for avoiding aliasing in the cycle and spectral frequency domains are obtained. Based on these findings, we propose algorithms for classifying BTSCLD, orthogonal frequency division multiplexing (OFDM), and SCLD signals, and for the blind estimation of the BT-SCLD block transmission parameters. Simulation and laboratory experiments demonstrate the effectiveness of the proposed algorithms under low signal-to-noise ratios (SNRs), short sensing times, and various channel conditions. Furthermore, these algorithms have the advantage of not requiring the recovery of carrier, waveform, and symbol timing information, or the estimation of signal and noise powers.

Cognitive radio (CR) systems require knowledge of the signal environment if they are to coexist with the primary (incumbent) users of the radio spectrum. Consequently, signal identification is a major task of a CR. This paper proposes a novel identification algorithm for spatial multiplexing (SM) and Alamouti space-time block code (AL STBC) transmissions, when the CR employs a single receive antenna. This algorithm relies on the fourth-order moment as a discriminating signal feature and uses a maximum likelihood (ML) criterion for decision making. Its performance is investigated through theoretical analysis and simulation experiments.