Search

Devices located in remote regions often lack coverage from well-developed terrestrial communication infrastructure. This not only prevents them from experiencing high quality communication services but also hinders the delivery of machine learning services in remote regions. In this paper, we propose a new federated learning (FL) methodology tailored to space-air-ground integrated networks (SAGINs) to tackle this issue. Our approach strategically leverages the nodes within space and air layers as both (i) edge computing units and (ii) model aggregators during the FL process, addressing the challenges that arise from the limited computation powers of ground devices and the absence of terrestrial base stations in the target region. The key idea behind our methodology is the adaptive data offloading and handover procedures that incorporate various network dynamics in SAGINs, including the mobility, heterogeneous computation powers, and inconsistent coverage times of incoming satellites. We analyze the latency of our scheme and develop an adaptive data offloading optimizer, and also characterize the theoretical convergence bound of our proposed algorithm. Experimental results confirm the advantage of our SAGIN-assisted FL methodology in terms of training time and test accuracy compared with various baselines., Comment: This paper is accepted for publication in IEEE Journal on Selected Areas in Communications (JSAC)

While network coverage maps continue to expand, many devices located in remote areas remain unconnected to terrestrial communication infrastructures, preventing them from getting access to the associated data-driven services. In this paper, we propose a ground-to-satellite cooperative federated learning (FL) methodology to facilitate machine learning service management over remote regions. Our methodology orchestrates satellite constellations to provide the following key functions during FL: (i) processing data offloaded from ground devices, (ii) aggregating models within device clusters, and (iii) relaying models/data to other satellites via inter-satellite links (ISLs). Due to the limited coverage time of each satellite over a particular remote area, we facilitate satellite transmission of trained models and acquired data to neighboring satellites via ISL, so that the incoming satellite can continue conducting FL for the region. We theoretically analyze the convergence behavior of our algorithm, and develop a training latency minimizer which optimizes over satellite-specific network resources, including the amount of data to be offloaded from ground devices to satellites and satellites' computation speeds. Through experiments on three datasets, we show that our methodology can significantly speed up the convergence of FL compared with terrestrial-only and other satellite baseline approaches., Comment: This paper is accepted for publication in IEEE Journal on Selected Areas in Communications (JSAC)

The conventional federated learning (FedL) architecture distributes machine learning (ML) across worker devices by having them train local models that are periodically aggregated by a server. FedL ignores two important characteristics of contemporary wireless networks, however: (i) the network may contain heterogeneous communication/computation resources, and (ii) there may be significant overlaps in devices' local data distributions. In this work, we develop a novel optimization methodology that jointly accounts for these factors via intelligent device sampling complemented by device-to-device (D2D) offloading. Our optimization methodology aims to select the best combination of sampled nodes and data offloading configuration to maximize FedL training accuracy while minimizing data processing and D2D communication resource consumption subject to realistic constraints on the network topology and device capabilities. Theoretical analysis of the D2D offloading subproblem leads to new FedL convergence bounds and an efficient sequential convex optimizer. Using these results, we develop a sampling methodology based on graph convolutional networks (GCNs) which learns the relationship between network attributes, sampled nodes, and D2D data offloading to maximize FedL accuracy. Through evaluation on popular datasets and real-world network measurements from our edge testbed, we find that our methodology outperforms popular device sampling methodologies from literature in terms of ML model performance, data processing overhead, and energy consumption., Comment: Published in IEEE/ACM Transactions on Networking. arXiv admin note: substantial text overlap with arXiv:2101.00787

The network edge's role in Artificial Intelligence (AI) inference processing is rapidly expanding, driven by a plethora of applications seeking computational advantages. These applications strive for data-driven efficiency, leveraging robust AI capabilities and prioritizing real-time responsiveness. However, as demand grows, so does system complexity. The proliferation of AI inference accelerators showcases innovation but also underscores challenges, particularly the varied software and hardware configurations of these devices. This diversity, while advantageous for certain tasks, introduces hurdles in device integration and coordination. In this paper, our objectives are three-fold. Firstly, we outline the requirements and components of a framework that accommodates hardware diversity. Next, we assess the impact of device heterogeneity on AI inference performance, identifying strategies to optimize outcomes without compromising service quality. Lastly, we shed light on the prevailing challenges and opportunities in this domain, offering insights for both the research community and industry stakeholders., Comment: This paper has been accepted for publication in the proceedings of the IEEE International Workshop on Computer Aided Modeling and Design of Communication Links and Networks 2023 (IEEE CAMAD 2023)

Federated learning (FL) has emerged as a key technique for distributed machine learning (ML). Most literature on FL has focused on ML model training for (i) a single task/model, with (ii) a synchronous scheme for updating model parameters, and (iii) a static data distribution setting across devices, which is often not realistic in practical wireless environments. To address this, we develop DMA-FL considering dynamic FL with multiple downstream tasks/models over an asynchronous model update architecture. We first characterize convergence via introducing scheduling tensors and rectangular functions to capture the impact of system parameters on learning performance. Our analysis sheds light on the joint impact of device training variables (e.g., number of local gradient descent steps), asynchronous scheduling decisions (i.e., when a device trains a task), and dynamic data drifts on the performance of ML training for different tasks. Leveraging these results, we formulate an optimization for jointly configuring resource allocation and device scheduling to strike an efficient trade-off between energy consumption and ML performance. Our solver for the resulting non-convex mixed integer program employs constraint relaxations and successive convex approximations with convergence guarantees. Through numerical experiments, we reveal that DMA-FL substantially improves the performance-efficiency tradeoff., Comment: Completed the major revision for IEEE Transactions on Cognitive Communications and Networking

Federated learning (FL) has been promoted as a popular technique for training machine learning (ML) models over edge/fog networks. Traditional implementations of FL have largely neglected the potential for inter-network cooperation, treating edge/fog devices and other infrastructure participating in ML as separate processing elements. Consequently, FL has been vulnerable to several dimensions of network heterogeneity, such as varying computation capabilities, communication resources, data qualities, and privacy demands. We advocate for cooperative federated learning (CFL), a cooperative edge/fog ML paradigm built on device-to-device (D2D) and device-to-server (D2S) interactions. Through D2D and D2S cooperation, CFL counteracts network heterogeneity in edge/fog networks through enabling a model/data/resource pooling mechanism, which will yield substantial improvements in ML model training quality and network resource consumption. We propose a set of core methodologies that form the foundation of D2D and D2S cooperation and present preliminary experiments that demonstrate their benefits. We also discuss new FL functionalities enabled by this cooperative framework such as the integration of unlabeled data and heterogeneous device privacy into ML model training. Finally, we describe some open research directions at the intersection of cooperative edge/fog and FL., Comment: This paper has been accepted for publication in IEEE Communications Magazine

Semi-decentralized federated learning blends the conventional device to-server (D2S) interaction structure of federated model training with localized device-to-device (D2D) communications. We study this architecture over practical edge networks with multiple D2D clusters modeled as time-varying and directed communication graphs. Our investigation results in an algorithm that controls the fundamental trade-off between (a) the rate of convergence of the model training process towards the global optimizer, and (b) the number of D2S transmissions required for global aggregation. Specifically, in our semi-decentralized methodology, D2D consensus updates are injected into the federated averaging framework based on column-stochastic weight matrices that encapsulate the connectivity within the clusters. To arrive at our algorithm, we show how the expected optimality gap in the current global model depends on the greatest two singular values of the weighted adjacency matrices (and hence on the densities) of the D2D clusters. We then derive tight bounds on these singular values in terms of the node degrees of the D2D clusters, and we use the resulting expressions to design a threshold on the number of clients required to participate in any given global aggregation round so as to ensure a desired convergence rate. Simulations performed on real-world datasets reveal that our connectivity-aware algorithm reduces the total communication cost required to reach a target accuracy significantly compared with baselines depending on the connectivity structure and the learning task., Comment: 10 pages, 5 figures. This paper has been accepted to ACM-MobiHoc 2023

As we increase the number of personal computing devices that we carry (mobile devices, tablets, e-readers, and laptops) and these come equipped with increasing resources, there is a vast potential computation power that can be utilized from those devices. Edge computing promises to exploit these underlying computation resources closer to users to help run latency-sensitive applications such as augmented reality and video analytics. However, one key missing piece has been how to incorporate personally owned unmanaged devices into a usable edge computing system. The primary challenges arise due to the heterogeneity, lack of interference management, and unpredictable availability of such devices. In this paper we propose an orchestration framework IBDASH, which orchestrates application tasks on an edge system that comprises a mix of commercial and personal edge devices. IBDASH targets reducing both end-to-end latency of execution and probability of failure for applications that have dependency among tasks, captured by directed acyclic graphs (DAGs). IBDASH takes memory constraints of each edge device and network bandwidth into consideration. To assess the effectiveness of IBDASH, we run real application tasks on real edge devices with widely varying capabilities.We feed these measurements into a simulator that runs IBDASH at scale. Compared to three state-of-the-art edge orchestration schemes, LAVEA, Petrel, and LaTS, and two intuitive baselines, IBDASH reduces the end-to-end latency and probability of failure, by 14% and 41% on average respectively. The main takeaway from our work is that it is feasible to combine personal and commercial devices into a usable edge computing platform, one that delivers low latency and predictable and high availability.

Symbol detection is a fundamental and challenging problem in modern communication systems, e.g., multiuser multiple-input multiple-output (MIMO) setting. Iterative Soft Interference Cancellation (SIC) is a state-of-the-art method for this task and recently motivated data-driven neural network models, e.g. DeepSIC, that can deal with unknown non-linear channels. However, these neural network models require thorough timeconsuming training of the networks before applying, and is thus not readily suitable for highly dynamic channels in practice. We introduce an online training framework that can swiftly adapt to any changes in the channel. Our proposed framework unifies the recent deep unfolding approaches with the emerging generative adversarial networks (GANs) to capture any changes in the channel and quickly adjust the networks to maintain the top performance of the model. We demonstrate that our framework significantly outperforms recent neural network models on highly dynamic channels and even surpasses those on the static channel in our experiments.

Federated learning (FL) has been recognized as one of the most promising solutions for distributed machine learning (ML). In most of the current literature, FL has been studied for supervised ML tasks, in which edge devices collect labeled data. Nevertheless, in many applications, it is impractical to assume existence of labeled data across devices. To this end, we develop a novel methodology, Cooperative Federated unsupervised Contrastive Learning (CF-CL), for FL across edge devices with unlabeled datasets. CF-CL employs local device cooperation where data are exchanged among devices through device-to-device (D2D) communications to avoid local model bias resulting from non-independent and identically distributed (non-i.i.d.) local datasets. CF-CL introduces a push-pull smart data sharing mechanism tailored to unsupervised FL settings, in which, each device pushes a subset of its local datapoints to its neighbors as reserved data points, and pulls a set of datapoints from its neighbors, sampled through a probabilistic importance sampling technique. We demonstrate that CF-CL leads to (i) alignment of unsupervised learned latent spaces across devices, (ii) faster global convergence, allowing for less frequent global model aggregations; and (iii) is effective in extreme non-i.i.d. data settings across the devices., Comment: Accepted for publication in IEEE Global Communications Conferences (GLOBECOM), 2022

We propose cooperative edge-assisted dynamic federated learning (CE-FL). CE-FL introduces a distributed machine learning (ML) architecture, where data collection is carried out at the end devices, while the model training is conducted cooperatively at the end devices and the edge servers, enabled via data offloading from the end devices to the edge servers through base stations. CE-FL also introduces floating aggregation point, where the local models generated at the devices and the servers are aggregated at an edge server, which varies from one model training round to another to cope with the network evolution in terms of data distribution and users' mobility. CE-FL considers the heterogeneity of network elements in terms of communication/computation models and the proximity to one another. CE-FL further presumes a dynamic environment with online variation of data at the network devices which causes a drift at the ML model performance. We model the processes taken during CE-FL, and conduct analytical convergence analysis of its ML model training. We then formulate network-aware CE-FL which aims to adaptively optimize all the network elements via tuning their contribution to the learning process, which turns out to be a non-convex mixed integer problem. Motivated by the large scale of the system, we propose a distributed optimization solver to break down the computation of the solution across the network elements. We finally demonstrate the effectiveness of our framework with the data collected from a real-world testbed.

Federated learning is a prime candidate for distributed machine learning at the network edge due to the low communication complexity and privacy protection among other attractive properties. However, existing algorithms face issues with slow convergence and/or robustness of performance due to the considerable heterogeneity of data distribution, computation and communication capability at the edge. In this work, we tackle both of these issues by focusing on the key component of model aggregation in federated learning systems and studying optimal algorithms to perform this task. Particularly, we propose a contextual aggregation scheme that achieves the optimal context-dependent bound on loss reduction in each round of optimization. The aforementioned context-dependent bound is derived from the particular participating devices in that round and an assumption on smoothness of the overall loss function. We show that this aggregation leads to a definite reduction of loss function at every round. Furthermore, we can integrate our aggregation with many existing algorithms to obtain the contextual versions. Our experimental results demonstrate significant improvements in convergence speed and robustness of the contextual versions compared to the original algorithms. We also consider different variants of the contextual aggregation and show robust performance even in the most extreme settings., Comment: 10 pages

Federated learning (FedL) has emerged as a popular technique for distributing model training over a set of wireless devices, via iterative local updates (at devices) and global aggregations (at the server). In this paper, we develop parallel successive learning (PSL), which expands the FedL architecture along three dimensions: (i) Network, allowing decentralized cooperation among the devices via device-to-device (D2D) communications. (ii) Heterogeneity, interpreted at three levels: (ii-a) Learning: PSL considers heterogeneous number of stochastic gradient descent iterations with different mini-batch sizes at the devices; (ii-b) Data: PSL presumes a dynamic environment with data arrival and departure, where the distributions of local datasets evolve over time, captured via a new metric for model/concept drift. (ii-c) Device: PSL considers devices with different computation and communication capabilities. (iii) Proximity, where devices have different distances to each other and the access point. PSL considers the realistic scenario where global aggregations are conducted with idle times in-between them for resource efficiency improvements, and incorporates data dispersion and model dispersion with local model condensation into FedL. Our analysis sheds light on the notion of cold vs. warmed up models, and model inertia in distributed machine learning. We then propose network-aware dynamic model tracking to optimize the model learning vs. resource efficiency tradeoff, which we show is an NP-hard signomial programming problem. We finally solve this problem through proposing a general optimization solver. Our numerical results reveal new findings on the interdependencies between the idle times in-between the global aggregations, model/concept drift, and D2D cooperation configuration.

Error correcting codes are a fundamental component in modern day communication systems, demanding extremely high throughput, ultra-reliability and low latency. Recent approaches using machine learning (ML) models as the decoders offer both improved performance and great adaptability to unknown environments, where traditional decoders struggle. We introduce a general framework to further boost the performance and applicability of ML models. We propose to combine ML decoders with a competing discriminator network that tries to distinguish between codewords and noisy words, and, hence, guides the decoding models to recover transmitted codewords. Our framework is game-theoretic, motivated by generative adversarial networks (GANs), with the decoder and discriminator competing in a zero-sum game. The decoder learns to simultaneously decode and generate codewords while the discriminator learns to tell the differences between decoded outputs and codewords. Thus, the decoder is able to decode noisy received signals into codewords, increasing the probability of successful decoding. We show a strong connection of our framework with the optimal maximum likelihood decoder by proving that this decoder defines a Nash equilibrium point of our game. Hence, training to equilibrium has a good possibility of achieving the optimal maximum likelihood performance. Moreover, our framework does not require training labels, which are typically unavailable during communications, and, thus, seemingly can be trained online and adapt to channel dynamics. To demonstrate the performance of our framework, we combine it with the very recent neural decoders and show improved performance compared to the original models and traditional decoding algorithms on various codes., Comment: 6 pages, accepted to GLOBECOM 2021

Edge computing has revolutionized the world of mobile and wireless networks world thanks to its flexible, secure, and performing characteristics. Lately, we have witnessed the increasing use of it to make more performing the deployment of machine learning (ML) techniques such as federated learning (FL). FL was debuted to improve communication efficiency compared to conventional distributed machine learning (ML). The original FL assumes a central aggregation server to aggregate locally optimized parameters and might bring reliability and latency issues. In this paper, we conduct an in-depth study of strategies to replace this central server by a flying master that is dynamically selected based on the current participants and/or available resources at every FL round of optimization. Specifically, we compare different metrics to select this flying master and assess consensus algorithms to perform the selection. Our results demonstrate a significant reduction of runtime using our flying master FL framework compared to the original FL from measurements results conducted in our EdgeAI testbed and over real 5G networks using an operational edge testbed., Comment: 6 pages, accepted to GLOBECOM 2021

We investigate training machine learning (ML) models across a set of geo-distributed, resource-constrained clusters of devices through unmanned aerial vehicles (UAV) swarms. The presence of time-varying data heterogeneity and computational resource inadequacy among device clusters motivate four key parts of our methodology: (i) stratified UAV swarms of leader, worker, and coordinator UAVs, (ii) hierarchical nested personalized federated learning (HN-PFL), a distributed ML framework for personalized model training across the worker-leader-core network hierarchy, (iii) cooperative UAV resource pooling to address computational inadequacy of devices by conducting model training among the UAV swarms, and (iv) model/concept drift to model time-varying data distributions. In doing so, we consider both micro (i.e., UAV-level) and macro (i.e., swarm-level) system design. At the micro-level, we propose network-aware HN-PFL, where we distributively orchestrate UAVs inside swarms to optimize energy consumption and ML model performance with performance guarantees. At the macro-level, we focus on swarm trajectory and learning duration design, which we formulate as a sequential decision making problem tackled via deep reinforcement learning. Our simulations demonstrate the improvements achieved by our methodology in terms of ML performance, network resource savings, and swarm trajectory efficiency., Comment: To be published in IEEE Transactions on Network and Service Management

While additional training data improves the robustness of deep neural networks against adversarial examples, it presents the challenge of curating a large number of specific real-world samples. We circumvent this challenge by using additional data from proxy distributions learned by advanced generative models. We first seek to formally understand the transfer of robustness from classifiers trained on proxy distributions to the real data distribution. We prove that the difference between the robustness of a classifier on the two distributions is upper bounded by the conditional Wasserstein distance between them. Next we use proxy distributions to significantly improve the performance of adversarial training on five different datasets. For example, we improve robust accuracy by up to 7.5% and 6.7% in $\ell_{\infty}$ and $\ell_2$ threat model over baselines that are not using proxy distributions on the CIFAR-10 dataset. We also improve certified robust accuracy by 7.6% on the CIFAR-10 dataset. We further demonstrate that different generative models bring a disparate improvement in the performance in robust training. We propose a robust discrimination approach to characterize the impact of individual generative models and further provide a deeper understanding of why current state-of-the-art in diffusion-based generative models are a better choice for proxy distribution than generative adversarial networks., Comment: ICLR 2022 version (30 pages, 13 figures, 12 tables)

We ask the following question: what training information is required to design an effective outlier/out-of-distribution (OOD) detector, i.e., detecting samples that lie far away from the training distribution? Since unlabeled data is easily accessible for many applications, the most compelling approach is to develop detectors based on only unlabeled in-distribution data. However, we observe that most existing detectors based on unlabeled data perform poorly, often equivalent to a random prediction. In contrast, existing state-of-the-art OOD detectors achieve impressive performance but require access to fine-grained data labels for supervised training. We propose SSD, an outlier detector based on only unlabeled in-distribution data. We use self-supervised representation learning followed by a Mahalanobis distance based detection in the feature space. We demonstrate that SSD outperforms most existing detectors based on unlabeled data by a large margin. Additionally, SSD even achieves performance on par, and sometimes even better, with supervised training based detectors. Finally, we expand our detection framework with two key extensions. First, we formulate few-shot OOD detection, in which the detector has access to only one to five samples from each class of the targeted OOD dataset. Second, we extend our framework to incorporate training data labels, if available. We find that our novel detection framework based on SSD displays enhanced performance with these extensions, and achieves state-of-the-art performance. Our code is publicly available at https://github.com/inspire-group/SSD., Comment: ICLR 2021

The conventional federated learning (FedL) architecture distributes machine learning (ML) across worker devices by having them train local models that are periodically aggregated by a server. FedL ignores two important characteristics of contemporary wireless networks, however: (i) the network may contain heterogeneous communication/computation resources, while (ii) there may be significant overlaps in devices' local data distributions. In this work, we develop a novel optimization methodology that jointly accounts for these factors via intelligent device sampling complemented by device-to-device (D2D) offloading. Our optimization aims to select the best combination of sampled nodes and data offloading configuration to maximize FedL training accuracy subject to realistic constraints on the network topology and device capabilities. Theoretical analysis of the D2D offloading subproblem leads to new FedL convergence bounds and an efficient sequential convex optimizer. Using this result, we develop a sampling methodology based on graph convolutional networks (GCNs) which learns the relationship between network attributes, sampled nodes, and resulting offloading that maximizes FedL accuracy. Through evaluation on real-world datasets and network measurements from our IoT testbed, we find that our methodology while sampling less than 5% of all devices outperforms conventional FedL substantially both in terms of trained model accuracy and required resource utilization., Comment: This paper is accepted for publication in the proceedings of 2021 IEEE International Conference on Computer Communications (INFOCOM)

In recent years, edge computing has become a popular choice for latency-sensitive applications like facial recognition and augmented reality because it is closer to the end users compared to the cloud. Although infrastructure providers are working toward creating managed edge networks, personal devices such as laptops and tablets, which are widely available and are underutilized, can also be used as potential edge devices. We call such devices Unmanaged Edge Devices (UEDs). Scheduling application tasks on such an unmanaged edge system is not straightforward because of three fundamental reasons-heterogeneity in the computational capacity of the UEDs, uncertainty in the availability of the UEDs (due to devices leaving the system), and interference among multiple tasks sharing a UED. In this paper, we present I-BOT, an interference-based orchestration scheme for latency-sensitive tasks on an Unmanaged Edge Platform (UEP). It minimizes the completion time of applications and is bandwidth efficient. I-BOT brings forth three innovations. First, it profiles and predicts the interference patterns of the tasks to make scheduling decisions. Second, it uses a feedback mechanism to adjust for changes in the computational capacity of the UEDs and a prediction mechanism to handle their sporadic exits. Third, it accounts for input dependence of tasks in its scheduling decision (such as, two tasks requiring the same input data). To evaluate I-BOT, we run end-to-end simulations with applications representing autonomous driving, composed of multiple tasks. We compare to two basic baselines (random and round-robin) and two state-of-the-arts, Lavea [SEC-2017] and Petrel [MSN-2018]. Compared to these baselines, I-BOT significantly reduces the average service time of application tasks. This reduction is more pronounced in dynamic heterogeneous environments, which would be the case in a UEP., Comment: 14 pages, 13 figures

As a research community, we are still lacking a systematic understanding of the progress on adversarial robustness which often makes it hard to identify the most promising ideas in training robust models. A key challenge in benchmarking robustness is that its evaluation is often error-prone leading to robustness overestimation. Our goal is to establish a standardized benchmark of adversarial robustness, which as accurately as possible reflects the robustness of the considered models within a reasonable computational budget. To this end, we start by considering the image classification task and introduce restrictions (possibly loosened in the future) on the allowed models. We evaluate adversarial robustness with AutoAttack, an ensemble of white- and black-box attacks, which was recently shown in a large-scale study to improve almost all robustness evaluations compared to the original publications. To prevent overadaptation of new defenses to AutoAttack, we welcome external evaluations based on adaptive attacks, especially where AutoAttack flags a potential overestimation of robustness. Our leaderboard, hosted at https://robustbench.github.io/, contains evaluations of 120+ models and aims at reflecting the current state of the art in image classification on a set of well-defined tasks in $\ell_\infty$- and $\ell_2$-threat models and on common corruptions, with possible extensions in the future. Additionally, we open-source the library https://github.com/RobustBench/robustbench that provides unified access to 80+ robust models to facilitate their downstream applications. Finally, based on the collected models, we analyze the impact of robustness on the performance on distribution shifts, calibration, out-of-distribution detection, fairness, privacy leakage, smoothness, and transferability., Comment: The camera-ready version accepted at the NeurIPS'21 Datasets and Benchmarks Track: 120+ evaluations, 80+ models, 7 leaderboards (Linf, L2, common corruptions; CIFAR-10, CIFAR-100, ImageNet), significantly expanded analysis part (calibration, fairness, privacy leakage, smoothness, transferability)

We study the effects of allowing paid prioritization arrangements in a market with content provider (CP) competition. We consider competing CPs who pay prioritization fees to a monopolistic ISP so as to offset the ISP's cost for investing in infrastructure to support fast lanes. Unlike prior works, our proposed model of users' content consumption accounts for multi-purchasing (i.e., users simultaneously subscribing to more than one CP). This model allows us to account for the "attention" received by each CP, and consequently to draw a contrast between how subscription-revenues and ad-revenues are impacted by paid prioritization. We show that there exist incentives for the ISP to build additional fast lanes subsidized by CPs with sufficiently high revenue (from either subscription fees or advertisements). We show that non-prioritized content providers need not lose users, yet may lose revenue from advertisements due to decreased attention from users. We further show that users will consume a wider variety of content in a prioritized regime, and that they can attain higher welfare provided that non-prioritized traffic is not throttled. We discuss some policy and practical implications of these findings and numerically validate them.

Federated learning has emerged recently as a promising solution for distributing machine learning tasks through modern networks of mobile devices. Recent studies have obtained lower bounds on the expected decrease in model loss that is achieved through each round of federated learning. However, convergence generally requires a large number of communication rounds, which induces delay in model training and is costly in terms of network resources. In this paper, we propose a fast-convergent federated learning algorithm, called FOLB, which performs intelligent sampling of devices in each round of model training to optimize the expected convergence speed. We first theoretically characterize a lower bound on improvement that can be obtained in each round if devices are selected according to the expected improvement their local models will provide to the current global model. Then, we show that FOLB obtains this bound through uniform sampling by weighting device updates according to their gradient information. FOLB is able to handle both communication and computation heterogeneity of devices by adapting the aggregations according to estimates of device's capabilities of contributing to the updates. We evaluate FOLB in comparison with existing federated learning algorithms and experimentally show its improvement in trained model accuracy, convergence speed, and/or model stability across various machine learning tasks and datasets., Comment: 17 pages

With increasing expressive power, deep neural networks have significantly improved the state-of-the-art on image classification datasets, such as ImageNet. In this paper, we investigate to what extent the increasing performance of deep neural networks is impacted by background features? In particular, we focus on background invariance, i.e., accuracy unaffected by switching background features and background influence, i.e., predictive power of background features itself when foreground is masked. We perform experiments with 32 different neural networks ranging from small-size networks to large-scale networks trained with up to one Billion images. Our investigations reveal that increasing expressive power of DNNs leads to higher influence of background features, while simultaneously, increases their ability to make the correct prediction when background features are removed or replaced with a randomly selected texture-based background., Comment: 6 pages, 5 figures

Machine learning (ML) tasks are becoming ubiquitous in today's network applications. Federated learning has emerged recently as a technique for training ML models at the network edge by leveraging processing capabilities across the nodes that collect the data. There are several challenges with employing conventional federated learning in contemporary networks, due to the significant heterogeneity in compute and communication capabilities that exist across devices. To address this, we advocate a new learning paradigm called fog learning which will intelligently distribute ML model training across the continuum of nodes from edge devices to cloud servers. Fog learning enhances federated learning along three major dimensions: network, heterogeneity, and proximity. It considers a multi-layer hybrid learning framework consisting of heterogeneous devices with various proximities. It accounts for the topology structures of the local networks among the heterogeneous nodes at each network layer, orchestrating them for collaborative/cooperative learning through device-to-device (D2D) communications. This migrates from star network topologies used for parameter transfers in federated learning to more distributed topologies at scale. We discuss several open research directions to realizing fog learning., Comment: This paper is accepted for publication in IEEE Communications Magazine

Attacks on Internet routing are typically viewed through the lens of availability and confidentiality, assuming an adversary that either discards traffic or performs eavesdropping. Yet, a strategic adversary can use routing attacks to compromise the security of critical Internet applications like Tor, certificate authorities, and the bitcoin network. In this paper, we survey such application-specific routing attacks and argue that both application-layer and network-layer defenses are essential and urgently needed. While application-layer defenses are easier to deploy in the short term, we hope that our work serves to provide much needed momentum for the deployment of network-layer defenses.

Cyberphysical systems (CPS) are ubiquitous in our personal and professional lives, and they promise to dramatically improve micro-communities (e.g., urban farms, hospitals), macro-communities (e.g., cities and metropolises), urban structures (e.g., smart homes and cars), and living structures (e.g., human bodies, synthetic genomes). The question that we address in this article pertains to designing these CPS systems to be resilient-from-the-ground-up, and through progressive learning, resilient-by-reaction. An optimally designed system is resilient to both unique attacks and recurrent attacks, the latter with a lower overhead. Overall, the notion of resilience can be thought of in the light of three main sources of lack of resilience, as follows: exogenous factors, such as natural variations and attack scenarios; mismatch between engineered designs and exogenous factors ranging from DDoS (distributed denial-of-service) attacks or other cybersecurity nightmares, so called "black swan" events, disabling critical services of the municipal electrical grids and other connected infrastructures, data breaches, and network failures; and the fragility of engineered designs themselves encompassing bugs, human-computer interactions (HCI), and the overall complexity of real-world systems. In the paper, our focus is on design and deployment innovations that are broadly applicable across a range of CPS application areas., Comment: 36 pages, 2 figures, NSF-supported workshop on Grand Challenges in Resilience, held at Purdue, March 20-21, 2019

Storage has become a constrained resource on smartphones. Gaming is a popular activity on mobile devices and the explosive growth in the number of games coupled with their growing size contributes to the storage crunch. Even where storage is plentiful, it takes a long time to download and install a heavy app before it can be launched. This paper presents AppStreamer, a novel technique for reducing the storage requirements or startup delay of mobile games, and heavy mobile apps in general. AppStreamer is based on the intuition that most apps do not need the entirety of its files (images, audio and video clips, etc.) at any one time. AppStreamer can, therefore, keep only a small part of the files on the device, akin to a "cache", and download the remainder from a cloud storage server or a nearby edge server when it predicts that the app will need them in the near future. AppStreamer continuously predicts file blocks for the near future as the user uses the app, and fetches them from the storage server before the user sees a stall due to missing resources. We implement AppStreamer at the Android file system layer. This ensures that the apps require no source code or modification, and the approach generalizes across apps. We evaluate AppStreamer using two popular games: Dead Effect 2, a 3D first-person shooter, and Fire Emblem Heroes, a 2D turn-based strategy role-playing game. Through a user study, 75% and 87% of the users respectively find that AppStreamer provides the same quality of user experience as the baseline where all files are stored on the device. AppStreamer cuts down the storage requirement by 87% for Dead Effect 2 and 86% for Fire Emblem Heroes., Comment: 12 pages; EWSN 2020

A large body of recent work has investigated the phenomenon of evasion attacks using adversarial examples for deep learning systems, where the addition of norm-bounded perturbations to the test inputs leads to incorrect output classification. Previous work has investigated this phenomenon in closed-world systems where training and test inputs follow a pre-specified distribution. However, real-world implementations of deep learning applications, such as autonomous driving and content classification are likely to operate in the open-world environment. In this paper, we demonstrate the success of open-world evasion attacks, where adversarial examples are generated from out-of-distribution inputs (OOD adversarial examples). In our study, we use 11 state-of-the-art neural network models trained on 3 image datasets of varying complexity. We first demonstrate that state-of-the-art detectors for out-of-distribution data are not robust against OOD adversarial examples. We then consider 5 known defenses for adversarial examples, including state-of-the-art robust training methods, and show that against these defenses, OOD adversarial examples can achieve up to 4$\times$ higher target success rates compared to adversarial examples generated from in-distribution data. We also take a quantitative look at how open-world evasion attacks may affect real-world systems. Finally, we present the first steps towards a robust open-world machine learning system., Comment: 18 pages, 5 figures, 9 tables

Fog computing, which distributes computing resources to multiple locations between the Internet of Things (IoT) devices and the cloud, is attracting considerable attention from academia and industry. Yet, despite the excitement about the potential of fog computing, few comprehensive quantitative characteristics of the properties of fog computing architectures have been conducted. In this paper we examine the properties of task completion latencies in fog computing. First, we present the results of our empirical benchmarking-based study of task completion latencies. The study covered a range of settings, and uniquely considered both traditional and serverless fog computing execution points. It demonstrated the range of execution point characteristics in different locations and the relative stability of latency characteristics for a given location. It also highlighted properties of serverless execution that are not incorporated in existing fog computing algorithms. Second, we present a framework we developed for co-optimizing task completion quality and latency, which was inspired by the insights of our empirical study. We describe fog computing task assignment problems we formulated under this framework, and present the algorithms we developed for solving them.

Social learning, i.e., students learning from each other through social interactions, has the potential to significantly scale up instruction in online education. In many cases, such as in massive open online courses (MOOCs), social learning is facilitated through discussion forums hosted by course providers. In this paper, we propose a probabilistic model for the process of learners posting on such forums, using point processes. Different from existing works, our method integrates topic modeling of the post text, timescale modeling of the decay in post activity over time, and learner topic interest modeling into a single model, and infers this information from user data. Our method also varies the excitation levels induced by posts according to the thread structure, to reflect typical notification settings in discussion forums. We experimentally validate the proposed model on three real-world MOOC datasets, with the largest one containing up to 6,000 learners making 40,000 posts in 5,000 threads. Results show that our model excels at thread recommendation, achieving significant improvement over a number of baselines, thus showing promise of being able to direct learners to threads that they are interested in more efficiently. Moreover, we demonstrate analytics that our model parameters can provide, such as the timescales of different topic categories in a course., Comment: To appear at ECML-PKDD 2018

The Rasch model is widely used for item response analysis in applications ranging from recommender systems to psychology, education, and finance. While a number of estimators have been proposed for the Rasch model over the last decades, the available analytical performance guarantees are mostly asymptotic. This paper provides a framework that relies on a novel linear minimum mean-squared error (L-MMSE) estimator which enables an exact, nonasymptotic, and closed-form analysis of the parameter estimation error under the Rasch model. The proposed framework provides guidelines on the number of items and responses required to attain low estimation errors in tests or surveys. We furthermore demonstrate its efficacy on a number of real-world collaborative filtering datasets, which reveals that the proposed L-MMSE estimator performs on par with state-of-the-art nonlinear estimators in terms of predictive performance., Comment: To be presented at ICML 2018

Sign recognition is an integral part of autonomous cars. Any misclassification of traffic signs can potentially lead to a multitude of disastrous consequences, ranging from a life-threatening accident to even a large-scale interruption of transportation services relying on autonomous cars. In this paper, we propose and examine security attacks against sign recognition systems for Deceiving Autonomous caRs with Toxic Signs (we call the proposed attacks DARTS). In particular, we introduce two novel methods to create these toxic signs. First, we propose Out-of-Distribution attacks, which expand the scope of adversarial examples by enabling the adversary to generate these starting from an arbitrary point in the image space compared to prior attacks which are restricted to existing training/test data (In-Distribution). Second, we present the Lenticular Printing attack, which relies on an optical phenomenon to deceive the traffic sign recognition system. We extensively evaluate the effectiveness of the proposed attacks in both virtual and real-world settings and consider both white-box and black-box threat models. Our results demonstrate that the proposed attacks are successful under both settings and threat models. We further show that Out-of-Distribution attacks can outperform In-Distribution attacks on classifiers defended using the adversarial training defense, exposing a new attack vector for these defenses., Comment: Submitted to ACM CCS 2018; Extended version of [1801.02780] Rogue Signs: Deceiving Traffic Sign Recognition with Malicious Ads and Logos

Co-flows model a modern scheduling setting that is commonly found in a variety of applications in distributed and cloud computing. A stochastic co-flow task contains a set of parallel flows with randomly distributed sizes. Further, many applications require non-preemptive scheduling of co-flow tasks. This paper gives an approximation algorithm for stochastic non-preemptive co-flow scheduling. The proposed approach uses a time-indexed linear relaxation, and uses its solution to come up with a feasible schedule. This algorithm is shown to achieve a competitive ratio of $(2\log{m}+1)(1+\sqrt{m}\Delta)(1+m{\Delta}){(3+\Delta)}/{2}$ for zero-release times, and $(2\log{m}+1)(1+\sqrt{m}\Delta)(1+m\Delta)(2+\Delta)$ for general release times, where $\Delta$ represents the upper bound of squared coefficient of variation of processing times, and $m$ is the number of servers., Comment: Some of the results have been fixed, mainly involving the CoV. The changes compared to the previous version are minor

Probit regression was first proposed by Bliss in 1934 to study mortality rates of insects. Since then, an extensive body of work has analyzed and used probit or related binary regression methods (such as logistic regression) in numerous applications and fields. This paper provides a fresh angle to such well-established binary regression methods. Concretely, we demonstrate that linearizing the probit model in combination with linear estimators performs on par with state-of-the-art nonlinear regression methods, such as posterior mean or maximum aposteriori estimation, for a broad range of real-world regression problems. We derive exact, closed-form, and nonasymptotic expressions for the mean-squared error of our linearized estimators, which clearly separates them from nonlinear regression methods that are typically difficult to analyze. We showcase the efficacy of our methods and results for a number of synthetic and real-world datasets, which demonstrates that linearized binary regression finds potential use in a variety of inference, estimation, signal processing, and machine learning applications that deal with binary-valued observations or measurements., Comment: To be presented at CISS (http://ee-ciss.princeton.edu/)

We propose a new real-world attack against the computer vision based systems of autonomous vehicles (AVs). Our novel Sign Embedding attack exploits the concept of adversarial examples to modify innocuous signs and advertisements in the environment such that they are classified as the adversary's desired traffic sign with high confidence. Our attack greatly expands the scope of the threat posed to AVs since adversaries are no longer restricted to just modifying existing traffic signs as in previous work. Our attack pipeline generates adversarial samples which are robust to the environmental conditions and noisy image transformations present in the physical world. We ensure this by including a variety of possible image transformations in the optimization problem used to generate adversarial samples. We verify the robustness of the adversarial samples by printing them out and carrying out drive-by tests simulating the conditions under which image capture would occur in a real-world scenario. We experimented with physical attack samples for different distances, lighting conditions and camera angles. In addition, extensive evaluations were carried out in the virtual setting for a variety of image transformations. The adversarial samples generated using our method have adversarial success rates in excess of 95% in the physical as well as virtual settings., Comment: Extended abstract accepted for the 1st Deep Learning and Security Workshop; 5 pages, 4 figures

Many recent proposals for anonymous communication omit from their security analyses a consideration of the effects of time on important system components. In practice, many components of anonymity systems, such as the client location and network structure, exhibit changes and patterns over time. In this paper, we focus on the effect of such temporal dynamics on the security of anonymity networks. We present Tempest, a suite of novel attacks based on (1) client mobility, (2) usage patterns, and (3) changes in the underlying network routing. Using experimental analysis on real-world datasets, we demonstrate that these temporal attacks degrade user privacy across a wide range of anonymity networks, including deployed systems such as Tor; path-selection protocols for Tor such as DeNASA, TAPS, and Counter-RAPTOR; and network-layer anonymity protocols for Internet routing such as Dovetail and HORNET. The degradation is in some cases surprisingly severe. For example, a single host failure or network route change could quickly and with high certainty identify the client's ISP to a malicious host or ISP. The adversary behind each attack is relatively weak - generally passive and in control of one network location or a small number of hosts. Our findings suggest that designers of anonymity systems should rigorously consider the impact of temporal dynamics when analyzing anonymity., Comment: Conference version to appear at The 18th Privacy Enhancing Technologies Symposium (PETS 2018)

Among storage components, hard disk drives (HDDs) have become the most commonly-used type of non-volatile storage due to their recent technological advances, including, enhanced energy efficacy and significantly-improved areal density. Such advances in HDDs have made them an inevitable part of numerous computing systems, including, personal computers, closed-circuit television (CCTV) systems, medical bedside monitors, and automated teller machines (ATMs). Despite the widespread use of HDDs and their critical role in real-world systems, there exist only a few research studies on the security of HDDs. In particular, prior research studies have discussed how HDDs can potentially leak critical private information through acoustic or electromagnetic emanations. Borrowing theoretical principles from acoustics and mechanics, we propose a novel denial-of-service (DoS) attack against HDDs that exploits a physical phenomenon, known as acoustic resonance. We perform a comprehensive examination of physical characteristics of several HDDs and create acoustic signals that cause significant vibrations in HDD's internal components. We demonstrate that such vibrations can negatively influence the performance of HDDs embedded in real-world systems. We show the feasibility of the proposed attack in two real-world case studies, namely, personal computers and CCTVs., Comment: 8 pages, 8 figures

This paper introduces an analytical framework to investigate optimal design choices for the placement of virtual controllers along the cloud-to-things continuum. The main application scenarios include low-latency cyber-physical systems in which real-time control actions are required in response to the changes in states of an IoT node. In such cases, deploying controller software on a cloud server is often not tolerable due to delay from the network edge to the cloud. Hence, it is desirable to trade reliability with latency by moving controller logic closer to the network edge. Modeling the IoT node as a dynamical system that evolves linearly in time with quadratic penalty for state deviations, recursive expressions for the optimum control policy and the resulting minimum cost value are obtained by taking virtual fog controller reliability and response time latency into account. Our results indicate that latency is more critical than reliability in provisoning virtualized control services over fog endpoints, as it determines the swiftness of the fog control system as well as the timeliness of state measurements. Based on a realistic drone trajectory tracking model, an extensive simulation study is also performed to illustrate the influence of reliability and latency on the control of autonomous vehicles over fog.

In recent years, numerous vehicular technologies, e.g., cruise control and steering assistant, have been proposed and deployed to improve the driving experience, passenger safety, and vehicle performance. Despite the existence of several novel vehicular applications in the literature, there still exists a significant gap between resources needed for a variety of vehicular (in particular, data-dominant, latency-sensitive, and computationally-heavy) applications and the capabilities of already-in-market vehicles. To address this gap, different smartphone-/Cloud-based approaches have been proposed that utilize the external computational/storage resources to enable new applications. However, their acceptance and application domain are still very limited due to programability, wireless connectivity, and performance limitations, along with several security/privacy concerns. In this paper, we present a novel architecture that can potentially enable rapid development of various vehicular applications while addressing shortcomings of smartphone-/Cloud-based approaches. The architecture is formed around a core component, called SmartCore, a privacy/security-friendly programmable dongle that brings general-purpose computational and storage resources to the vehicle and hosts in-vehicle applications. Based on the proposed architecture, we develop an application development framework for vehicles, that we call ProCMotive. ProCMotive enables developers to build customized vehicular applications along the Cloud-to-edge continuum, i.e., different functions of an application can be distributed across SmartCore, the user's personal devices, and the Cloud. To highlight potential benefits that the framework provides, we design and develop two different vehicular applications based on ProCMotive, namely, Amber Response and Insurance Monitor., Comment: 23 Pages

Tor is vulnerable to network-level adversaries who can observe both ends of the communication to deanonymize users. Recent work has shown that Tor is susceptible to the previously unknown active BGP routing attacks, called RAPTOR attacks, which expose Tor users to more network-level adversaries. In this paper, we aim to mitigate and detect such active routing attacks against Tor. First, we present a new measurement study on the resilience of the Tor network to active BGP prefix attacks. We show that ASes with high Tor bandwidth can be less resilient to attacks than other ASes. Second, we present a new Tor guard relay selection algorithm that incorporates resilience of relays into consideration to proactively mitigate such attacks. We show that the algorithm successfully improves the security for Tor clients by up to 36% on average (up to 166% for certain clients). Finally, we build a live BGP monitoring system that can detect routing anomalies on the Tor network in real time by performing an AS origin check and novel detection analytics. Our monitoring system successfully detects simulated attacks that are modeled after multiple known attack types as well as a real-world hijack attack (performed by us), while having low false positive rates., Comment: Appearing at IEEE S&P 2017

To cope with recent exponential increases in demand for mobile data, wireless Internet service providers (ISPs) are increasingly changing their pricing plans and deploying WiFi hotspots to offload their mobile traffic. However, these ISP-centric approaches for traffic management do not always match the interests of mobile users. Users face a complex, multi-dimensional tradeoff between cost, throughput, and delay in making their offloading decisions: while they may save money and receive a higher throughput by waiting for WiFi access, they may not wait for WiFi if they are sensitive to delay. To navigate this tradeoff, we develop AMUSE (Adaptive bandwidth Management through USer-Empowerment), a functional prototype of a practical, cost-aware WiFi offloading system that takes into account a user's throughput-delay tradeoffs and cellular budget constraint. Based on predicted future usage and WiFi availability, AMUSE decides which applications to offload to what times of the day. Since nearly all traffic flows from mobile devices are TCP flows, we introduce a new receiver-side bandwidth allocation mechanism to practically enforce the assigned rate of each TCP application. Thus, AMUSE users can optimize their bandwidth rates according to their own cost-throughput-delay tradeoff without relying on support from different apps' content servers. Through a measurement study of 20 smartphone users' traffic usage traces, we observe that though users already offload a large amount of some application types, our framework can offload a significant additional portion of users' cellular traffic. We implement AMUSE and evaluate its effectiveness with 3G and WiFi usage data obtained from a trial with 37 mobile users. Our results show that AMUSE improves user utility, when compared with AMUSE, other offloading algorithms yield 14\% and 27\% lower user utilities for light and heavy users, respectively., Comment: 15 pages, 16 figures, IEEE Transactions on Mobile Computing, Vol. 15, No. 5, May 2016

The past 15 years have seen the rise of the Cloud, along with rapid increase in Internet backbone traffic and more sophisticated cellular core networks. There are three different types of Clouds: (1) data center, (2) backbone IP network and (3) cellular core network, responsible for computation, storage, communication and network management. Now the functions of these three types of Clouds are descending to be among or near the end users, i.e., to the edge of networks, as Fog. This article presents an overview on research opportunities of Fog networking: an architecture that users one or a collaborative multitude of end-user clients or near-user edge devices to carry out a substantial amount of storage, communication and management. Architecture allocates functionalities, while engineering artifacts that may use a Fog architecture include 5G, home/personal networking, and the Internet of Things.

We study student behavior and performance in two Massive Open Online Courses (MOOCs). In doing so, we present two frameworks by which video-watching clickstreams can be represented: one based on the sequence of events created, and another on the sequence of positions visited. With the event-based framework, we extract recurring subsequences of student behavior, which contain fundamental characteris- tics such as reflecting (i.e., repeatedly playing and pausing) and revising (i.e., plays and skip backs). We find that some of these behaviors are significantly associated with whether a user will be Correct on First Attempt (CFA) or not in answering quiz questions. With the position-based framework, we then devise models for performance. In evaluating these through CFA prediction, we find that three of them can substantially improve prediction quality in terms of accuracy and F1, which underlines the ability to relate behavior to performance. Since our prediction considers videos individually, these benefits also suggest that our models are useful in situations where there is limited training data, e.g., for early detection or in short courses.