arXiv subjects

Osvaldo Simeone

Publications and source records attributed to Osvaldo Simeone.

At least 253 records · Page 14Linked to original sources

arXiv2020

Decentralized Federated Learning via SGD over Wireless D2D Networks

Federated Learning (FL), an emerging paradigm for fast intelligent acquisition at the network edge, enables joint training of a machine learning model over distributed data sets and computing resources with limited disclosure of local data. Communication is a critical enabler of large-scale FL due to significant amount of model information exchanged among edge devices. In this paper, we consider a network of wireless devices sharing a common fading wireless channel for the deployment of FL. Each device holds a generally distinct training set, and communication typically takes place in a Device-to-Device (D2D) manner. In the ideal case in which all devices within communication range can communicate simultaneously and noiselessly, a standard protocol that is guaranteed to converge to an optimal solution of the global empirical risk minimization problem under convexity and connectivity assumptions is Decentralized Stochastic Gradient Descent (DSGD). DSGD integrates local SGD steps with periodic consensus averages that require communication between neighboring devices. In this paper, wireless protocols are proposed that implement DSGD by accounting for the presence of path loss, fading, blockages, and mutual interference. The proposed protocols are based on graph coloring for scheduling and on both digital and analog transmission strategies at the physical layer, with the latter leveraging over-the-air computing via sparsity-based recovery.

cs.IT↗

arXiv2020

Cooperative Learning via Federated Distillation over Fading Channels

Cooperative training methods for distributed machine learning are typically based on the exchange of local gradients or local model parameters. The latter approach is known as Federated Learning (FL). An alternative solution with reduced communication overhead, referred to as Federated Distillation (FD), was recently proposed that exchanges only averaged model outputs. While prior work studied implementations of FL over wireless fading channels, here we propose wireless protocols for FD and for an enhanced version thereof that leverages an offline communication phase to communicate ``mixed-up'' covariate vectors. The proposed implementations consist of different combinations of digital schemes based on separate source-channel coding and of over-the-air computing strategies based on analog joint source-channel coding. It is shown that the enhanced version FD has the potential to significantly outperform FL in the presence of limited spectral resources.

eess.SP↗

arXiv2020

Inter-Tenant Cooperative Reception for C-RAN Systems With Spectrum Pooling

This work studies the uplink of a multi-tenant cloud radio access network (C-RAN) system with spectrum pooling. In the system, each operator has a cloud processor (CP) connected to a set of proprietary radio units (RUs) through finite-capacity fronthaul links. The uplink spectrum is divided into private and shared subbands, and all the user equipments (UEs) of the participating operators can simultaneously transmit signals on the shared subband. To mitigate inter-operator interference on the shared subband, the CPs of the participating operators can exchange compressed uplink baseband signals on finite-capacity backhaul links. This work tackles the problem of jointly optimizing bandwidth allocation, transmit power control and fronthaul compression strategies. In the optimization, we impose that the inter-operator privacy loss be limited by a given threshold value. An iterative algorithm is proposed to find a suboptimal solution based on the matrix fractional programming approach. Numerical results validate the advantages of the proposed optimized spectrum pooling scheme.

eess.SP↗

arXiv2020

Information-Centric Grant-Free Access for IoT Fog Networks: Edge vs Cloud Detection and Learning

A multi-cell Fog-Radio Access Network (F-RAN) architecture is considered in which Internet of Things (IoT) devices periodically make noisy observations of a Quantity of Interest (QoI) and transmit using grant-free access in the uplink. The devices in each cell are connected to an Edge Node (EN), which may also have a finite-capacity fronthaul link to a central processor. In contrast to conventional information-agnostic protocols, the devices transmit using a Type-Based Multiple Access (TBMA) protocol that is tailored to enable the estimate of the field of correlated QoIs in each cell based on the measurements received from IoT devices. In this paper, this form of information-centric radio access is studied for the first time in a multi-cell F-RAN model with edge or cloud detection. Edge and cloud detection are designed and compared for a multi-cell system. Optimal model-based detectors are introduced and the resulting asymptotic behavior of the probability of error at cloud and edge is derived. Then, for the scenario in which a statistical model is not available, data-driven edge and cloud detectors are discussed and evaluated in numerical results.

cs.IT↗

arXiv2020

ITENE: Intrinsic Transfer Entropy Neural Estimator

Quantifying the directionality of information flow is instrumental in understanding, and possibly controlling, the operation of many complex systems, such as transportation, social, neural, or gene-regulatory networks. The standard Transfer Entropy (TE) metric follows Granger's causality principle by measuring the Mutual Information (MI) between the past states of a source signal $X$ and the future state of a target signal $Y$ while conditioning on past states of $Y$. Hence, the TE quantifies the improvement, as measured by the log-loss, in the prediction of the target sequence $Y$ that can be accrued when, in addition to the past of $Y$, one also has available past samples from $X$. However, by conditioning on the past of $Y$, the TE also measures information that can be synergistically extracted by observing both the past of $X$ and $Y$, and not solely the past of $X$. Building on a private key agreement formulation, the Intrinsic TE (ITE) aims to discount such synergistic information to quantify the degree to which $X$ is \emph{individually} predictive of $Y$, independent of $Y$'s past. In this paper, an estimator of the ITE is proposed that is inspired by the recently proposed Mutual Information Neural Estimation (MINE). The estimator is based on variational bound on the KL divergence, two-sample neural network classifiers, and the pathwise estimator of Monte Carlo gradients.

cs.IT↗

arXiv2020

Fundamental Limits of Wireless Caching under Uneven-Capacity Channels

This work identifies the fundamental limits of cache-aided coded multicasting in the presence of the well-known `worst-user' bottleneck. This stems from the presence of receiving users with uneven channel capacities, which often forces the rate of transmission of each multicasting message to be reduced to that of the slowest user. This bottleneck, which can be detrimental in general wireless broadcast settings, motivates the analysis of coded caching over a standard Single-Input-Single-Output (SISO) Broadcast Channel (BC) with K cache-aided receivers, each with a generally different channel capacity. For this setting, we design a communication algorithm that is based on superposition coding that capitalizes on the realization that the user with the worst channel may not be the real bottleneck of communication. We then proceed to provide a converse that shows the algorithm to be near optimal, identifying the fundamental limits of this setting within a multiplicative factor of 4. Interestingly, the result reveals that, even if several users are experiencing channels with reduced capacity, the system can achieve the same optimal delivery time that would be achievable if all users enjoyed maximal capacity.

cs.IT↗

arXiv2020

From Learning to Meta-Learning: Reduced Training Overhead and Complexity for Communication Systems

Machine learning methods adapt the parameters of a model, constrained to lie in a given model class, by using a fixed learning procedure based on data or active observations. Adaptation is done on a per-task basis, and retraining is needed when the system configuration changes. The resulting inefficiency in terms of data and training time requirements can be mitigated, if domain knowledge is available, by selecting a suitable model class and learning procedure, collectively known as inductive bias. However, it is generally difficult to encode prior knowledge into an inductive bias, particularly with black-box model classes such as neural networks. Meta-learning provides a way to automatize the selection of an inductive bias. Meta-learning leverages data or active observations from tasks that are expected to be related to future, and a priori unknown, tasks of interest. With a meta-trained inductive bias, training of a machine learning model can be potentially carried out with reduced training data and/or time complexity. This paper provides a high-level introduction to meta-learning with applications to communication systems.

cs.LG↗

arXiv2019

Non-Orthogonal eMBB-URLLC Radio Access for Cloud Radio Access Networks with Analog Fronthauling

This paper considers the coexistence of Ultra Reliable Low Latency Communications (URLLC) and enhanced Mobile BroadBand (eMBB) services in the uplink of Cloud Radio Access Network (C-RAN) architecture based on the relaying of radio signals over analog fronthaul links. While Orthogonal Multiple Access (OMA) to the radio resources enables the isolation and the separate design of different 5G services, Non-Orthogonal Multiple Access (NOMA) can enhance the system performance by sharing wireless and fronthaul resources. This paper provides an information-theoretic perspective in the performance of URLLC and eMBB traffic under both OMA and NOMA. The analysis focuses on standard cellular models with additive Gaussian noise links and a finite inter-cell interference span, and it accounts for different decoding strategies such as puncturing, Treating Interference as Noise (TIN) and Successive Interference Cancellation (SIC). Numerical results demonstrate that, for the considered analog fronthauling C-RAN architecture, NOMA achieves higher eMBB rates with respect to OMA, while guaranteeing reliable low-rate URLLC communication with minimal access latency. Moreover, NOMA under SIC is seen to achieve the best performance, while, unlike the case with digital capacity-constrained fronthaul links, TIN always outperforms puncturing.

cs.IT↗

arXiv2019

Online Reinforcement Learning of X-Haul Content Delivery Mode in Fog Radio Access Networks

We consider a Fog Radio Access Network (F-RAN) with a Base Band Unit (BBU) in the cloud and multiple cache-enabled enhanced Remote Radio Heads (eRRHs). The system aims at delivering contents on demand with minimal average latency from a time-varying library of popular contents. Information about uncached requested files can be transferred from the cloud to the eRRHs by following either backhaul or fronthaul modes. The backhaul mode transfers fractions of the requested files, while the fronthaul mode transmits quantized baseband samples as in Cloud-RAN (C-RAN). The backhaul mode allows the caches of the eRRHs to be updated, which may lower future delivery latencies. In contrast, the fronthaul mode enables cooperative C-RAN transmissions that may reduce the current delivery latency. Taking into account the trade-off between current and future delivery performance, this paper proposes an adaptive selection method between the two delivery modes to minimize the long-term delivery latency. Assuming an unknown and time-varying popularity model, the method is based on model-free Reinforcement Learning (RL). Numerical results confirm the effectiveness of the proposed RL scheme.

eess.SP↗

arXiv2019

Cache-Aided Content Delivery in Fog-RAN Systems with Topological Information and no CSI

In this work, we consider a Fog Radio Access Network (F-RAN) system with a partially connected wireless topology and no channel state information available at the cloud and Edge Nodes (ENs). An F-RAN consists of ENs that are equipped with caches and connected to the cloud through fronthaul links. We first focus on the case where cloud connectivity is disabled, and hence the ENs have to satisfy the user demands based only on their cache contents. For a class of partially connected regular networks, we present a delivery scheme which combines intra-file MDS coded caching at the ENs and blind interference avoidance on the wireless edge channel. This scheme requires minimum storage and leads to an achievable Normalized Delivery Time (NDT) that is within a constant multiplicative factor of the best known NDT with full storage. We next provide achievable schemes for the case where cloud connectivity is enabled, and we provide new insights on the interplay between cloud connectivity and edge caching when only topological information is available.

cs.IT↗

arXiv2019

Training Dynamic Exponential Family Models with Causal and Lateral Dependencies for Generalized Neuromorphic Computing

Neuromorphic hardware platforms, such as Intel's Loihi chip, support the implementation of Spiking Neural Networks (SNNs) as an energy-efficient alternative to Artificial Neural Networks (ANNs). SNNs are networks of neurons with internal analogue dynamics that communicate by means of binary time series. In this work, a probabilistic model is introduced for a generalized set-up in which the synaptic time series can take values in an arbitrary alphabet and are characterized by both causal and instantaneous statistical dependencies. The model, which can be considered as an extension of exponential family harmoniums to time series, is introduced by means of a hybrid directed-undirected graphical representation. Furthermore, distributed learning rules are derived for Maximum Likelihood and Bayesian criteria under the assumption of fully observed time series in the training set.

cs.LG↗

arXiv2019

Private and Secure Distributed Matrix Multiplication with Flexible Communication Load

Large matrix multiplications are central to large-scale machine learning applications. These operations are often carried out on a distributed computing platform with a master server and multiple workers in the cloud operating in parallel. For such distributed platforms, it has been recently shown that coding over the input data matrices can reduce the computational delay, yielding a trade-off between recovery threshold, i.e., the number of workers required to recover the matrix product, and communication load, i.e., the total amount of data to be downloaded from the workers. In this paper, in addition to exact recovery requirements, we impose security and privacy constraints on the data matrices, and study the recovery threshold as a function of the communication load. We first assume that both matrices contain private information and that workers can collude to eavesdrop on the content of these data matrices. For this problem, we introduce a novel class of secure codes, referred to as secure generalized PolyDot (SGPD) codes, that generalize state-of-the-art non-secure codes for matrix multiplication. SGPD codes allow a flexible trade-off between recovery threshold and communication load for a fixed maximum number of colluding workers while providing perfect secrecy for the two data matrices. We then study a connection between secure matrix multiplication and private information retrieval. We specifically assume that one of the data matrices is taken from a public set known to all the workers. In this setup, the identity of the matrix of interest should be kept private from the workers. For this model, we present a variant of generalized PolyDot codes that can guarantee both secrecy of one matrix and privacy for the identity of the other matrix for the case of no colluding servers.

cs.IT↗

arXiv2019

Optimizing Pilots and Analog Processing for Channel Estimation in Cell-Free Massive MIMO With One-Bit ADCs

In a cell-free cloud radio access network (C-RAN) architecture, uplink channel estimation is carried out by a centralized baseband processing unit (BBU) connected to distributed remote radio heads (RRHs). When the RRHs have multiple antennas and limited radio front-end resources, the design of uplink channel estimation is faced with the challenges posed by reduced radio frequency (RF) chains and one-bit analog-to-digital converters (ADCs) at the RRHs. This work tackles the problem of jointly optimizing the pilot sequences and the pre-RF chains analog combiners with the goal of minimizing the sum of mean squared errors (MSEs) of the estimated channel vectors at the BBU. The problem formulation models the impact of the ADC operation by leveraging Bussgang's theorem. An efficient solution is developed by means of an iterative alternating optimization algorithm. Numerical results validate the advantages of the proposed joint design compared to baseline schemes that randomly choose either pilots or analog combiners.

eess.SP↗

arXiv2019

Robust Baseband Compression Against Congestion in Packet-Based Fronthaul Networks Using Multiple Description Coding

In modern implementations of Cloud Radio Access Network (C-RAN), the fronthaul transport network will often be packet-based and it will have a multi-hop architecture built with general-purpose switches using network function virtualization (NFV) and software-defined networking (SDN). This paper studies the joint design of uplink radio and fronthaul transmission strategies for a C-RAN with a packet-based fronthaul network. To make an efficient use of multiple routes that carry fronthaul packets from remote radio heads (RRHs) to cloud, as an alternative to more conventional packet-based multiroute reception or coding, a multiple description coding (MDC) strategy is introduced that operates directly at the level of baseband signals. MDC ensures an improved quality of the signal received at the cloud in conditions of low network congestion, i.e., when more fronthaul packets are received within a tolerated deadline. The advantages of the proposed MDC approach as compared to the traditional path diversity scheme are validated via extensive numerical results.

cs.IT↗

arXiv2019

Information-Theoretic Analysis of D2D-Aided Pipelined Content Delivery in Fog-RAN

In a Fog-Radio Access Network (F-RAN), edge caching and fronthaul connectivity to a cloud processor are utilized for the purpose of content delivery. Additional Device-to-Device (D2D) communication capabilities can support the operation of an F-RAN by alleviating fronthaul and cloud processing load, and reducing the delivery time. In this work, basic limits on the normalized delivery time (NDT) metric, which captures the high signal-to-noise ratio worst-case latency for delivering any requested content to the users, are derived. Assuming proactive offline caching, out-of-band D2D communication, and an F-RAN with two edge nodes and two users, an information-theoretically optimal caching and delivery strategy is presented. Unlike prior work, the NDT performance is studied under pipelined transmission, whereby the edge nodes transmit on the wireless channel while simultaneously receiving messages over the fronthaul links, and the users transmit messages over the D2D links while at the same time receiving on the wireless channel. Insights are provided on the regimes in which D2D communication is beneficial, and the maximum improvement to the latency is characterized.

cs.IT↗

arXiv2019

Fog-Aided Wireless Networks for Content Delivery: Fundamental Latency Trade-Offs

A fog-aided wireless network architecture is studied in which edge-nodes (ENs), such as base stations, are connected to a cloud processor via dedicated fronthaul links, while also being endowed with caches. Cloud processing enables the centralized implementation of cooperative transmission strategies at the ENs, albeit at the cost of an increased latency due to fronthaul transfer. In contrast, the proactive caching of popular content at the ENs allows for the low-latency delivery of the cached files, but with generally limited opportunities for cooperative transmission among the ENs. The interplay between cloud processing and edge caching is addressed from an information-theoretic viewpoint by investigating the fundamental limits of a high Signal-to-Noise-Ratio (SNR) metric, termed normalized delivery time (NDT), which captures the worst-case coding latency for delivering any requested content to the users. The NDT is defined under the assumptions of either serial or pipelined fronthaul-edge transmission, and is studied as a function of fronthaul and cache capacity constraints. Placement and delivery strategies across both fronthaul and wireless, or edge, segments are proposed with the aim of minimizing the NDT. Information-theoretic lower bounds on the NDT are also derived. Achievability arguments and lower bounds are leveraged to characterize the minimal NDT in a number of important special cases, including systems with no caching capabilities, as well as to prove that the proposed schemes achieve optimality within a constant multiplicative factor of 2 for all values of the problem parameters.

cs.IT↗

arXiv2019

Zero-Delay Source-Channel Coding with a One-Bit ADC Front End and Correlated Side Information at the Receiver

Zero-delay transmission of a Gaussian source over an additive white Gaussian noise (AWGN) channel is considered with a one-bit analog-to-digital converter (ADC) front end and a correlated side information at the receiver. The design of the optimal encoder and decoder is studied for two performance criteria, namely, the mean squared error (MSE) distortion and the distortion outage probability (DOP), under an average power constraint on the channel input. For both criteria, necessary optimality conditions for the encoder and the decoder are derived. Using these conditions, it is observed that the numerically optimized encoder (NOE) under the MSE distortion criterion is periodic, and its period increases with the correlation between the source and the receiver side information. For the DOP, it is instead seen that the NOE mappings periodically acquire positive and negative values, which decay to zero with increasing source magnitude, and the interval over which the mapping takes non-zero values, becomes wider with the correlation between the source and the side information.

cs.IT↗

arXiv2019

End-to-end Learning of Waveform Generation and Detection for Radar Systems

An end-to-end learning approach is proposed for the joint design of transmitted waveform and detector in a radar system. Detector and transmitted waveform are trained alternately: For a fixed transmitted waveform, the detector is trained using supervised learning so as to approximate the Neyman-Pearson detector; and for a fixed detector, the transmitted waveform is trained using reinforcement learning based on feedback from the receiver. No prior knowledge is assumed about the target and clutter models. Both transmitter and receiver are implemented as feedforward neural networks. Numerical results show that the proposed end-to-end learning approach is able to obtain a more robust radar performance in clutter and colored noise of arbitrary probability density functions as compared to conventional methods, and to successfully adapt the transmitted waveform to environmental conditions.

eess.SP↗