Effective service deployment and orchestration will be essential to accommodate user workloads with diverse requirements in cloud-native beyond 5G Radio Access Networks (RAN). Orchestration will have to take into account individual service quality requirements, latency constraints, and dependencies, while leveraging unique characteristics of dominant workloads, such as machine learning (ML) models. In this work, we address the orchestration of ML-based services, considering users that request application services that rely on network services, such as localization, positioning, etc. Each service is composed of functions, at potentially different quality levels. The objective is to maximize the network operator’s revenue by determining service deployment, quality selection and computational resource allocation. The resulting problem is a mixed-integer non-convex problem, which we show is NP-hard. We provide sufficient conditions for the problem to be submodular, and for the general case we propose JADES, which relies on linear relaxation and convexification to decompose the problem into two subproblems, which are solved iteratively until convergence, followed by dependent randomized rounding. Our evaluation based on synthetic workloads shows that JADES outperforms baselines in terms of operator revenue and computational efficiency.

Edge computing(EC) is emerging as a key enabler for latency-sensitive applications such asAugmented Reality(AR), autonomous driving, and industrial IoT, by bringing computational resources closer toWireless Devices(WDs). However, the limited computational capacity inherent to EC presents challenges in resource allocation and in designing pricing mechanisms that provide the right incentives and are aligned with the user-perceived service quality. This paper addresses these challenges by formulating a Stackelberg game that models WDs’ valuation of EC services based on their offloading rates and the service quality they receive. We prove the existence of Stackelberg equilibria and we propose a tractable approximation technique based on log-barrier functions for computing approximate equilibria. Furthermore, to overcome computational issues, we build on the concept of a Differential Stackelberg Equilibrium (DSE) and we propose Stackelberg Gradient Play (SGP), an implicit gradient-based algorithm that ensures convergence to DSE while maintaining efficiency. Extensive simulations show that our approach significantly outperforms existing methods, achieving up to 70% higher revenue for the edge operator while reducing computational overhead substantially. These results underscore the viability of our framework for use in EC systems that require fast, adaptive, and service-aware joint resource management and pricing.

We consider correlated equilibria in an adversarial environment, where an adversary can compromise the public signal used by the players for choosing their strategies, while players aim at detecting a potential attack as soon as possible to avoid loss of utility. We model the interaction between the adversary and the players as a zero-sum game and we derive the maxmin strategies for both the defender and the attacker using the framework of quickest change detection. We define a class of adversarial strategies that achieve the optimal trade-off between the impact and the detectability of the attack for the adversary and show that a generalized CUSUM scheme is asymptotically optimal for their detection. Our numerical results on the Sioux-Falls benchmark traffic routing game show that the proposed detection scheme can effectively limit the utility loss by a potential adversary.

Adversarial patches are capable of misleading computer vision systems based on convolutional neural networks. Existing recovery methods suffer of at least one of three fundamental shortcomings: no information about the presence of patches in the scene, inability to efficiently handle noncontiguous patch attacks, and a strong reliance on fixed saliency thresholds. We propose Saliuitl, a recovery method independent of the number of patches and their shape, which unlike prior works, explicitly detects patch attacks before attempting recovery. In our approach, detection is based on the attributes of a binarized feature map ensemble, which is generated by using an ensemble of saliency thresholds. If an attack is detected, Saliuitl recovers clean predictions locating patches guided by an ensemble of binarized feature maps and inpainting them. We evaluate Saliuitl on widely used object detection and image classification benchmarks from the adversarial patch literature, and our results show that compared to recent state-of-the-art defenses, Saliuitl achieves a recovery rate up to 97.81 and 42.63 percentage points higher at the same rate of lost predictions for image classification and object detection, respectively. By design, Saliuitl has low computational complexity and is robust to adaptive white-box attacks. Our code is available at https://github.com/Saliuitl/Saliuitl/tree/main.

We address the problem of detecting adversarial attacks against cooperative multi-agent reinforcement learning with continuous action space. We propose a decentralized detector that relies solely on the local observations of the agents and makes use of a statistical characterization of the normal behavior of observable agents. The proposed detector utilizes deep neural networks to approximate the normal behavior of agents as parametric multivariate Gaussian distributions. Based on the predicted density functions, we define a normality score and provide a characterization of its mean and variance. This characterization allows us to employ a two-sided CUSUM procedure for detecting deviations of the normality score from its mean, serving as a detector of anomalous behavior in real-time. We evaluate our scheme on various multi-agent PettingZoo benchmarks against different state-of-the-art attack methods, and our results demonstrate the effectiveness of our method in detecting impactful adversarial attacks. Particularly, it outperforms the discrete counterpart by achieving AUC-ROC scores of over 0.95 against the most impactful attacks in all evaluated environments.

We consider a Network Operator (NO) that owns Edge Computing (EC) resources, virtualizes them and lets third party Service Providers (SPs) run their services, using the allocated slice of resources. We focus on one specific resource, i.e., cache space, and on the problem of how to allocate it among several SPs in order to minimize the backhaul traffic. Due to confidentiality guarantees, the NO cannot observe the nature of the traffic of SPs, which is encrypted. Allocation decisions are thus challenging, since they must be taken solely based on observed monitoring information. Another challenge is that not all the traffic is cacheable. We propose a data-driven cache allocation strategy, based on Reinforcement Learning (RL). Unlike most RL applications, in which the decision policy is learned offline on a simulator, we assume no previous knowledge is available to build such a simulator. We thus apply RL in anonlinefashion, i.e., the model and the policy are learned by directly perturbing and monitoring the actual system. Since perturbations generate spurious traffic, we thus need to limit perturbations. This requires learning to be extremely efficient. To this aim, we devise a strategy that learns an approximation of the cost function, while interacting with the system. We then use such an approximation in a Model-Based RL (MB-RL) to speed up convergence. We prove analytically that our strategy brings cache allocation boundedly close to the optimum and stably remains in such an allocation. We show in simulations that such convergence is obtained within few minutes. We also study its fairness, its sensitivity to several scenario characteristics and compare it with a method from the state-of-the-art.

Real-time cyber situational awareness (SA) is crucial for effective and timely incident response. However, maintaining SA requires substantial human effort; security analysts must analyze large volumes of alerts, many of which are false positives triggered by anomaly-based intrusion detection systems (IDSs). Efficiently prioritizing these alerts is vital to enable analysts to focus on real threats without delay. In this paper, we present two key contributions designed to improve real-time SA. First, we propose modeling dynamic alert prioritization as an active learning problem in a hidden Markov model (HMM) with the objective to minimize the mean squared error (MSE) of the belief. We propose to use the uncertainty of the belief as a proxy for the MSE of the belief, and we develop two computationally tractable policies for choosing alerts to investigate. Second, we propose and evaluate a state space and an exploit space reduction method to reduce the computational complexity of the belief update. We use simulations on synthetic and real dependency graphs to evaluate the proposed policies. Our results show that the proposed investigation policies reduce the MSE of the belief by up to 50% compared to baseline policies, and they are robust to high false alert rates and to investigation errors. Our results also show that state space reduction can reduce the computation time by 85% without a significant increase in the belief MSE.

Multi-access edge computing (MEC)-enabled integrated space-air-ground (SAG) networks have drawn much attention recently, as they can provide communication and computing services to wireless devices in areas that lack terrestrial base stations (TBSs). Leveraging the ample bandwidth in the terahertz (THz) spectrum, in this paper, we propose MEC-enabled integrated SAG networks with collaboration among unmanned aerial vehicles (UAVs). We then formulate the problem of minimizing the energy consumption of devices and UAVs in the proposed MEC-enabled integrated SAG networks by optimizing tasks offloading decisions, THz sub-bands assignment, transmit power control, and UAVs deployment. The formulated problem is a mixed-integer nonlinear programming (MILP) problem with a non-convex structure, which is challenging to solve. We thus propose a block coordinate descent (BCD) approach to decompose the problem into four sub-problems: 1) device task offloading decision problem, 2) THz sub-band assignment and power control problem, 3) UAV deployment problem, and 4) UAV task offloading decision problem. We then propose to use a matching game, concave-convex procedure (CCP) method, successive convex approximation (SCA), and block successive upper-bound minimization (BSUM) approaches for solving the individual subproblems. Finally, extensive simulations are performed to demonstrate the effectiveness of our proposed algorithm.

The commercial adoption of Edge Computing (EC) will require pricing schemes that cater to the financial interests of the operators and of the users. Pricing in EC is particularly challenging as it has to take into account the limited amount of edge resources as well as the stochasticity of user workloads due to location-specific workload characteristics and differences in user activity. We formulate the problem of maximizing the revenue of a serverless edge operator through dynamically pricing compute and memory resources under time varying workloads as a sequential decision making problem under uncertainty. We provide analytical results for the optimal pricing strategy in a Markovian setting in steady state. For the general case, we propose a novel Generalized Hidden Parameter Markov Decision Process (GHP-MDP) formulation of the revenue maximization problem, and we propose a dual Bayesian neural network approximator as a solution. The key novelty of the proposed solution is that it can be pre-trained on synthetic traces and adapts fast to previously unseen workload characteristics. We use simulations based on synthetic and real traffic traces to show that the proposed solution is sample-efficient thanks to effective transfer learning, and it outperforms state-of-the-art learning approaches in terms of revenue and learning rate by up to 50% on real traces.

As cyberattacks are becoming more sophisticated, automated activity logging and anomaly detection are becoming important tools for defending computer systems. Recent deep learning-based approaches have demonstrated promising results in cybersecurity contexts, typically using supervised learning combined with large amounts of labeled data. Self-supervised learning has seen growing interest as a method of training models because it does not require labeled training data, which can be difficult and expensive to collect. However, existing self-supervised approaches to anomaly detection in user authentication logs either suffer from low precision or rely on large pre-trained natural language models. This makes them slow and expensive both during training and inference. Building on previous works, we therefore propose an end-to-end trained self-supervised transformer-based sequence model for anomaly detection in user authentication events. Thanks in part to an adapted masked-language modeling (MLM) learning task and domain knowledge-based improvements to the anomaly detection method, our proposed model outperforms previous long short-term memory (LSTM)-based approaches at detecting red-team activity in the "Comprehensive, Multi-Source Cyber-Security Events" authentication event dataset, improving the area under the receiver operating characteristic curve (AUC) from 0.9760 to 0.9989 and achieving an average precision of 0.0410. Our work presents the first application of end-to-end trained self-supervised transformer models to user authentication data in a cybersecurity context, and demonstrates the potential of transformer-based approaches for anomaly detection.

Timely detection of cyber attacks is essential for minimizing attack impact, but it requires accurate real-time situational awareness (SA). In practice, SA is hampered by frequent false alerts from anomaly-based intrusion detection systems (IDS), causing alarm fatigue. Investigating alerts by humans can enhance SA, but it is resource-intensive and it is often unclear which alerts to prioritize. In this paper, we propose a framework for optimizing human-in-the-loop attack detection, consisting of three key components: 1) dynamic alert prioritization, which ranks alerts based on previous alerts and investigations, 2) human alert investigation, referring to the manual analysis of alerts, and 3) sequential hypothesis testing, a method that confirms a hypothesis based on incoming alerts, with pruned hidden Markov models (HMMs). We formulate the problem as that of active learning in an HMM, and we propose two alert prioritization policies, namely Max Ratio and Max KL. The proposed policies aim to select the most informative alerts based on historical data and prior investigations, thereby minimizing the detection time. Simulation results show that our proposed policies reduce the time to detection by up to 79% compared to a static baseline policy, while maintaining a target mean time between false detections (MTBFD).

Motivated by the emergence of function-as-a-service (FaaS) as a programming abstraction for edge computing, we consider the problem of caching and pricing applications for edge computation offloading in a dynamic environment whereWirelesss Devices(WDs) can be active or inactive at any point in time. We model the problem as a single leader multiple-follower Stackelberg game, where the service operator is the leader and decides what applications to cache and how much to charge for their use, while the WDs are the followers and decide whether or not to offload their computations. We show that the WDs' interaction can be modeled as a player-specific congestion game and show the existence and computability of equilibria. We then show that under perfect and complete information the equilibrium price of the service operator can be computed in polynomial time for any cache placement. For the incomplete information case, we propose a Bayesian Gaussian Process Bandit algorithm for learning an optimal price for a cache placement and provide a bound on its asymptotic regret. We then propose a Gaussian process approximation-based greedy heuristic for computing the cache placement. We use extensive simulations to evaluate the proposed learning scheme, and show that it outperforms state of the art algorithms by up to 50% at little computational overhead.

We consider the problem of resource allocation, pricing and application caching for latency sensitive task of floading in serverless edge computing. We model the interaction between a profit-maximizing operator and cost-minimizing Wireless Devices (WDs) as a Stackelberg game where the operator is the leader and decides the price, resource allocation and set of applications to cache, while the WDs are the followers and decide whether to offload their tasks. We first show that the game has a Subgame Perfect Equilibrium (SPE), but computing it, is NP-hard. Importantly, we show that an SPE, which maximizes the operator's revenue, results in minimal energy consumption among the WDs. For computing an approximate SPE, we propose a linear time approximation algorithm with bounded approximation ratio for resource allocation and pricing, and we propose an efficient heuristic based on the utility density of individual applications for the joint optimization of caching, resource allocation and pricing. Our results show that the proposed algorithm outperforms state-of-the-art methods by up to an order of magnitude both in terms of revenue and total energy savings and has small computational overhead. An interesting feature of our results is that the utility of the operator is maximized by a solution that maximizes the WDs' energy savings through computation offloading, which makes it a promising candidate for energy efficient edge cloud deployments.

Caching plays a central role in networked systems, reducing the load on servers and the delay experienced by users. Despite their relevance, networked caching systems still pose a number of challenges pertaining their long term behavior. In this paper, we formally show and experimentally evidence conditions under which networked caches tend to synchronize over time. Such synchronization, in turn, leads to performance degradation and aging, motivating the monitoring of caching systems for eventual rejuvenation, as well as the deployment of diverse cache replacement policies across caches to promote diversity and preclude synchronization and its aging effects. Based on trace-driven simulations with real workloads, we show how hit probability is sensitive to varying channel reliability, cache sizes, and cache separation, indicating that the mix of simple policies, such as Least Recently Used (LRU) and Least Frequently Used (LFU), provide competitive performance against state-of-art policies. Indeed, our results suggest that diversity in cache replacement policies, rejuvenation and intentional dropping of requests are strategies that build diversity across caches, preventing or mitigating performance degradation due to caching aging.

The increasing demand for latency-sensitive applications has necessitated the development of sophisticated algorithms that efficiently manage packets with end-to-end delay targets traversing the networked infrastructure. Network components must consider minimizing the packets' end-to-end delay violation probabilities (DVP) as a guiding principle throughout the transmission path to ensure timely deliveries. Active queue management (AQM) schemes are commonly used to mitigate congestion by dropping packets and controlling queuing delay. Today's established AQM schemes are threshold-driven, identifying congestion and trigger packet dropping using a predefined criteria which is unaware of packets' DVPs. In this work, we propose a novel framework, Delta, that combines end-to-end delay characterization with AQM for minimizing DVP. In a queuing theoretic environment, we show that such a policy is feasible by utilizing a data-driven approach to predict the queued packets' DVPs. That enables Delta AQM to effectively handle links with arbitrary stationary service time processes. The implementation is described in detail, and its performance is evaluated and compared with state of the art AQM algorithms. Our results show the Delta outperforms current AQM schemes substantially, in particular in scenarios where high reliability, i.e. high quantiles of the tail latency distribution, are of interest.

Integrated space-air-ground (ISAG) networks have emerged as a promising technology for next-generation commu-nication networks, which demand higher throughput and wider coverage. However, several challenges must be addressed, such as wireless resource and interference management, transmit power control, and optimal deployment of unmanned aerial vehicles (UAV s) to achieve higher throughput. To this end, in this paper, we propose the joint transmit power control and the deployment UAV problem in the ISAG network to maximize the sum-rate of devices while guaranteeing the minimum rate requirement of each device and the wireless backhaul link constraints between UAV s and the satellite. However, solving the resulting problem is challenging since it has a non-convex structure. As a solution, we propose to decompose the problem into two subproblems. We then transform the decomposed subproblems into convex forms and solve them using successive convex approximation (SCA). Finally, we conduct extensive simulations to show the effectiveness of the proposed method, and the numerical results show that the proposed method achieves a performance gain: up to 8.35%, and 4.08% in comparison to FPA and C-UAVs schemes.

We consider the problem of detecting adversarial attacks against cooperative multi-agent reinforcement learning. We propose a decentralized scheme that allows agents to detect the abnormal behavior of one compromised agent. Our approach is based on a recurrent neural network (RNN) trained during cooperative learning to predict the action distribution of other agents based on local observations. The predicted distribution is used for computing a normality score for the agents, which allows the detection of the misbehavior of other agents. To explore the robustness of the proposed detection scheme, we formulate the worst-case attack against our scheme as a constrained reinforcement learning problem. We propose to compute an attack policy by optimizing the corresponding dual function using reinforcement learning. Extensive simulations on various multi-agent benchmarks show the effectiveness of the proposed detection scheme in detecting state-of-the-art attacks and in limiting the impact of undetectable attacks.

We consider the interplay between latency constrained applications and function-level resource management in a serverless edge computing environment. We develop a game theoretic model of the interaction between rate adaptive applications and a load balancing operator under a function-oriented pay-as-you-go pricing model. We show that under perfect information, the strategic interaction between the applications can be formulated as a generalized Nash equilibrium problem, and use variational inequality theory to prove that the game admits an equilibrium. For the case of imperfect information, we propose an online learning algorithm for applications to maximize their utility through rate adaptation and resource reservation. We show that the proposed algorithm can converge to equilibria and achieves zero regret asymptotically, and our simulation results show that the algorithm achieves good system performance at equilibrium, ensures fast convergence, and enables applications to meet their latency constraints.

Mobile edge computing (MEC) is an emerging architecture for accommodating latency sensitive virtualized services (VSs). Many of these VSs are expected to be safety critical, and will have some form of reliability requirements. In order to support provisioning reliability to such VSs in MEC in an efficient and confidentiality preserving manner, in this paper we consider the joint resource dimensioning and placement problem for VSs with diverse reliability requirements, with the objective of minimizing the energy consumption. We formulate the problem as an integer programming problem, and prove that it is NP-hard. We propose a two-step approximation algorithm with bounded approximation ratio based on Lagrangian relaxation. We benchmark our algorithm against two greedy algorithms in realistic scenarios. The results show that the proposed solution is computationally efficient, scalable and can provide up to 30 percent reduction in energy consumption compared to greedy algorithms.

Network slicing is a promising approach for enabling low latency computation offloading in edge computing systems. In this paper, we consider an edge computing system under network slicing in which the wireless devices generate latency sensitive computational tasks. We address the problem of joint dynamic assignment of computational tasks to slices, management of radio resources across slices and management of radio and computing resources within slices. We formulate theJoint Slice Selection and Edge Resource Management(JSS-ERM) problem as a mixed-integer problem with the objective to minimize the completion time of computational tasks. We show that the JSS-ERM problem is NP-hard and develop an approximation algorithm with bounded approximation ratio based on a game theoretic treatment of the problem. We use extensive simulations to provide insight into the performance of the proposed solution from the perspective of the whole system and from the perspective of individual slices. Our results show that the proposed slicing policy can achieve significant gains compared to the equal slicing policy, and that the computational complexity of the proposed task placement algorithm is approximately linear in the number of devices.

5G and beyond 5G low power wireless networks make Internet of Things (IoT) and Cyber-Physical Systems (CPS) applications capable of serving massive amounts of devices and machines. Due to the broadcast nature of wireless networks, it is crucial to secure the communication between these devices and machines from spoofing and interception attacks. This paper is concerned with the security of carrier frequency offset (CFO) based continuous physical layer authentication. The interaction between an attacker and a defender is modeled as a dynamic discrete leader-follower game with imperfect information. In the considered model, a legitimate user (Alice) communicates with the defender/operator (Bob) and is authorized by her CFO continuously. The attacker (Eve), by listening/eavesdropping the communication between Alice and Bob, tries to learn the CFO characteristics of Alice and aims to inject malicious packets to Bob by impersonating Alice. First, by showing that the optimal attacker strategy is a threshold policy, an optimization problem of the attacker with exponentially growing action space is reduced to a tractable integer optimization problem with a single parameter, then the corresponding defender cost is derived. Extensive simulations illustrate the characteristics of optimal strategies/utilities of the players depending on the actions, and show that the defender’s optimal false positive rate causes attack success probabilities to be in the order of 0.99. The results show the importance of the parameters while finding the balance between system security and efficiency.

We consider the computation offloading problem in an edge computing system in which an operator jointly manages wireless and computing resources across devices that make their offloading decisions autonomously with the objective to minimize their own completion times. We develop a game theoretical model of the interaction between the devices and an operator that can implement one of two resource allocation policies, a cost minimizing or a time fair resource allocation policy. We express the optimal cost minimizing resource allocation policy in closed form and prove the existence of Stackelberg equilibria for both resource allocation policies. We propose two efficient decentralized algorithms that devices can use for computing equilibria of offloading decisions under the cost minimizing and the time fair resource allocation policies. We establish bounds on the price of anarchy of the games played by the devices and by doing so we show that the proposed algorithms have bounded approximation ratios. Our simulation results show that the cost minimizing resource allocation policy can achieve significantly lower completion times than the time fair allocation policy. At the same time, the convergence time of the proposed algorithms is approximately linear in the number of devices, and thus they could be effectively implemented for edge computing resource management.

Efficient communication and computing resource allocation is becoming a fundamental issue in wireless networks. Efficiency is most often defined in terms of throughput, utilization and spectral efficiency, while the required computational effort is often overlooked. In this paper, we focus on efficient and computationally lightweight downlink scheduling, and we propose a semi-persistent scheduler based on adaptive short term traffic prediction. We evaluate the performance of the proposed scheduler in terms of throughput, fairness, latency, and scheduling complexity. Our numerical results show that scheduling with prediction is a promising approach in improving network performance. The proposed semi-persistent scheduler performs equally well in terms of throughput, fairness, and latency as traditional proportional-fair scheduling, but at a significantly reduced computational cost.

Network traffic prediction is a fundamental prerequisite for dynamic resource provisioning in wireline and wireless networks, but is known to be challenging due to non-stationarity and due to its burstiness and self-similar nature. The prediction of network traffic at the user level is particularly challenging, because the traffic characteristics emerge from a complex interaction of user level and application protocol behavior. In this work we address the problem of predicting the network traffic at the user level over a short horizon, motivated by its applications in cellular scheduling. Motivated by recent works on robust adversarial learning, we treat the prediction problem for non-stationary traffic in an adversarial context, and propose a meta-learning scheme that consists of a set of predictors, each optimized to predict a particular kind of traffic, and of a master policy that is trained for choosing the best fit predictor dynamically based on recent prediction performance, using deep reinforcement learning. We evaluate the proposed meta-learning scheme on a variety of traffic traces consisting of video and non-video traffic. Our results show that it consistently outperforms state-of-the-art predictors, and can adapt to before unseen traffic without the need for retraining the individual predictors.

Precise time synchronization of Phasor Measurement Units (PMUs) is critical for monitoring and control of smart grids. Thus, time synchronization attacks (TSAs) against PMUs pose a severe threat to smart grid security. In this paper we present an approach for detecting TSAs based on the interaction between the time synchronization system and the power system. We develop a phasor measurement model and use it to derive an accurate closed form expression for the correlation between the frequency adjustments made by the PMU clock and the resulting change in the measured phase angle, without an attack. We then propose one model-based and three data-driven TSA detectors that exploit the change in correlation due to a TSA. Using extensive simulations, we evaluate the proposed detectors under different strategies for implementing TSAs, and show that the proposed detectors are superior to state-of-the-art clock frequency anomaly detection, especially for unstable clocks.

Motivated by various delay sensitive applications, we address the problem of coordinating the offloading decisions of wireless devices that periodically generate computationally intensive tasks. We consider autonomous devices that aim at minimizing their own cost by choosing when to perform their tasks and whether or not to offload their tasks to an edge cloud through one of the multiple wireless links. We develop a game theoretical model of the problem, prove the existence of pure strategy Nash equilibria and propose a polynomial complexity algorithm for computing an equilibrium. Furthermore, we characterize the structure of the equilibria, and by providing an upper bound on the price of anarchy of the game we establish an asymptotically tight bound on the approximation ratio of the proposed algorithm. Our simulation results show that the proposed algorithm achieves significant performance gain compared to uncoordinated computation offloading at a computational complexity that is on average linear in the number of devices.

We consider the problem of allocating wireless and computing resources to a set of autonomous wireless devices in an edge computing system. Devices in the system can decide whether or not to use edge computing resources for offloading computing tasks so as to minimize their completion time, while the edge cloud operator can allocate wireless and computing resources to the devices. We model the interaction between devices and the operator as a Stackelberg game, prove the existence of Stackelberg equilibria, and propose an efficient decentralized algorithm for computing equilibria. We provide a bound on the price of anarchy of the game, which also serves as an approximation ratio bound for the proposed algorithm. Our simulation results show that the joint allocation of wireless and computing resources by the operator can halve the completion times compared to a system with static resource allocation. At the same time, the convergence time of the proposed algorithm is approximately linear in the number of devices, and thus it could be effectively implemented for edge computing resource management.

Offloading computation to a mobile cloud is a promising solution to augment the computation capabilities of mobile devices. In this paper, we consider selfish mobile devices in a dense wireless network, in which individual mobile devices can offload computations through multiple access points or through the base station to a mobile cloud so as to minimize their computation costs. We provide a game theoretical analysis of the problem, prove the existence of pure strategy Nash equilibria, and provide an efficient decentralized algorithm for computing an equilibrium. For the case when the cloud computing resources scale with the number of mobile devices, we show that all improvement paths are finite. Furthermore, we provide an upper bound on the price of anarchy of the game, which serves as an upper bound on the approximation ratio of the proposed decentralized algorithms. We use simulations to evaluate the time complexity of computing Nash equilibria and to provide insights into the price of anarchy of the game under realistic scenarios. Our results show that the equilibrium cost may be close to optimal, and the convergence time is almost linear in the number of mobile devices.

Age of information has been recently proposed to quantify the freshness of information, e.g., in cyber-physical systems, where it is of critical importance. Motivated by wireless camera networks where multi-view image processing is required, in this paper we propose to extend the concept of age of information to capture packets carrying correlated data. We consider a system consisting of wireless camera nodes with overlapping fields of view and a set of processing nodes, and address the problem of the joint optimization of processing node assignment and camera transmission scheduling, so as to minimize the maximum peak age of information from all sources. We formulate the multi-view age minimization (MVAM) problem, and prove its NP-hardness under the two widely used interference models as well as with given candidate transmitting groups. We provide fundamental results including tractable cases and optimality conditions of the MVAM problem for two baseline scenarios. To solve MVAM efficiently, we develop an optimization algorithm based on a decomposition approach. Numerical results show that by employing our approach the maximum peak age is significantly reduced in comparison to a traditional centralized solution with minimum-time scheduling.

Fog computing is identified as a key enabler for using various emerging applications by battery powered and computationally constrained devices. In this paper, we consider devices that aim at improving their performance by choosing to offload their computational tasks to nearby devices or to an edge cloud. We develop a game theoretical model of the problem and use a variational inequality theory to compute an equilibrium task allocation in static mixed strategies. Based on the computed equilibrium strategy, we develop a decentralized algorithm for allocating the computational tasks among nearby devices and the edge cloud. We use the extensive simulations to provide insight into the performance of the proposed algorithm and compare its performance with the performance of a myopic best response algorithm that requires global knowledge of the system state. Despite the fact that the proposed algorithm relies on average system parameters only, our results show that it provides a good system performance close to that of the myopic best response algorithm.

We consider the uplink of a single cell multi-user multiple input multiple output (MU-MIMO) system, in which the base station acquires channel state information at the receiver by means of uplink pilot signals. Since each mobile station has a sum power budget that is used to transmit pilot and data symbols, the pilot power ratio (PPR) has a large impact on the system performance in terms of spectral and energy efficiency. We formulate the problem of PPR setting as a non-cooperative game, in which each mobile station aims at minimizing the mean squared error of the uplink received data symbols at the base station. We show that in this game a unique Nash equilibrium exists, and propose an iterative decentralized algorithm-termed best PPR algorithm (BPA)-that is guaranteed to converge to that Nash equilibrium. Since BPA dynamically responds to the measured interference, it outperforms widely used schemes that use a predetermined PPR. BPA also performs close to the global optimum, especially when mobile stations with similar path loss values are co-scheduled in the MU-MIMO system. Based on these insights, we propose a practical signaling mechanism for implementing BPA in MU-MIMO systems.

Real-time visual analysis tasks, like tracking and recognition, require swift execution of computationally intensive algorithms. Visual sensor networks could be enabled to perform such tasks by allowing the camera nodes to offload their computational load to nearby processing nodes. In this paper, we address the problem of minimizing the completion time of multiple camera sensors that share the transmission and the processing resources of multiple processing nodes for computation offloading. We show that the problem is NP-hard, and propose a combination of central coordination and distributed optimization with limited signaling among the camera sensors as a solution. We analyze the existence of equilibrium allocations for the distributed algorithms, evaluate the effect of the network topology and of the video characteristics on the algorithms' performance, and assess the benefits of central coordination. Our results demonstrate that with sufficient information available, distributed optimization can provide low completion times, moreover predictable and stable performance can be achieved with additional, sparse central coordination.

Motivated by the proliferation of wireless building automation systems (BAS) and increasing security-awareness among BAS operators, in this paper, we propose a taxonomy for the security assessment of BASs. We apply the proposed taxonomy to Thread, an emerging native IP-based protocol for BAS. Our analysis reveals a number of potential weaknesses in the design of Thread. We propose potential solutions for mitigating several identified weaknesses and discuss their efficacy. We also provide suggestions for improvements in future versions of the standard. Overall, our analysis shows that Thread has a well-designed security control for the targeted use case, making it a promising candidate for communication in next generation BASs.

Replacing hardware controllers with software-based virtual process control functions (VPFs) is a promising approach for improving the operational efficiency and flexibility of industrial control systems. VPFs can be executed in edge clouds in 5G mobile networks or in the wireless backhaul, which can further improve efficiency. Nonetheless, for the acceptance of virtualization in industrial control systems, a fundamental challenge is to ensure that the placement of VPFs be resilient to component failures and cyber-attacks, besides being efficient. In this paper we address this challenge by considering that VPF placement costs are incurred by reserving mobile edge computing (MEC) resources, executing VPF instances, and by data communication. We formulate the VPF placement problem as an integer programming problem, considering resilience as a constraint. We propose a solution based on generalized Benders decomposition and based on linear relaxation of the resulting sub-problems, which effectively reduces the number of integer variables to the number of MEC nodes. We evaluate the proposed solution with respect to operational cost, efficiency, and scalability in a simulated metropolitan area. Our results show that the proposed solution reduces the total cost significantly compared to a greedy baseline algorithm and a local search heuristic, and can scale to moderate problem instances.

In-network caching is an appealing solution to cope with the increasing bandwidth demand of video, audio, and data transfer over the Internet. Nonetheless, in order to protect consumer privacy and their own business, content providers (CPs) increasingly deliver encrypted content, thereby preventing Internet service providers (ISPs) from employing traditional caching strategies, which require the knowledge of the objects being transmitted. To overcome this emerging tussle between security and efficiency, in this paper we propose an architecture in which the ISP partitions the cache space into slices, assigns each slice to a different CP, and lets the CPs remotely manage their slices. This architecture enables transparent caching of encrypted content and can be deployed in the very edge of the ISP's network (i.e., base stations and femtocells), while allowing CPs to maintain exclusive control over their content. We propose an algorithm, called SDCP, for partitioning the cache storage into slices so as to maximize the bandwidth savings provided by the cache. A distinctive feature of our algorithm is that ISPs only need to measure the aggregated miss rates of each CP, but they need not know the individual objects that are requested. We prove that the SDCP algorithm converges to a partitioning that is close to the optimal, and we bound its optimality gap. We use simulations to evaluate SDCP's convergence rate under stationary and nonstationary content popularity. Finally, we show that SDCP significantly outperforms traditional reactive caching techniques, considering both CPs with perfect and with imperfect knowledge of their content popularity.

Various industrial control applications have stringent end-to-end latency requirements in the order of a few milliseconds. Software-defined networking (SDN) is a promising solution in order to meet these stringent requirements under varying traffic patterns, as it enables the flexible management of flows across the network. Thus, SDN allows to ensure that traffic flows use congestion-free paths, reducing the delay to forwarding and processing delays at the SDN nodes. However, accommodating new flows at runtime is under such a setting challenging as it may require the migration of existing flows, without interrupting ongoing traffic. In this paper, we consider the problem of dynamic flow migration and propose a polynomial time algorithm that can find a solution if direct flow migration is feasible. We furthermore propose an algorithm for computing both direct and indirect flow migration and prove its correctness. Numerical results obtained on a FatTree network topology show that flow migration is typically necessary for networks with a moderate number of flows, while direct flow migration is feasible in around 60% of the cases.

Offloading computation to a mobile cloud is a promising approach for enabling the use of computationally intensive applications by mobile devices. In this paper we consider autonomous devices that maximize their own performance by choosing one of many wireless access points for computation offloading. We develop a game theoretic model of the problem, prove the existence of pure strategy Nash equilibria, and provide a polynomial time algorithm for computing an equilibrium. For the case when the cloud computing resources scale with the number of mobile devices we show that all improvement paths are finite. We provide a bound on the price of anarchy of the game, thus our algorithm serves as an approximation algorithm for the global computation offloading cost minimization problem. We use extensive simulations to provide insight into the performance and the convergence time of the algorithms in various scenarios. Our results show that the equilibrium cost may be close to optimal, and the convergence time is almost linear in the number of mobile devices.

In this paper, we consider the problem of optimal dynamic service function (SF) placement and flow routing in a SF chaining (SFC) enabled network. We formulate a multi-objective optimization problem to maximize the acceptable flow rate and to minimize the energy cost for multiple service chains. We transform the multi-objective optimization problem into a single-objective mixed integer linear programming (MILP) problem, and prove that the problem is NP-hard. We propose a polynomial time algorithm based on linear relaxation and rounding to approximate the optimal solution of the MILP. Extensive simulations are conducted to evaluate the effects of the energy budget, the network topology, and the amount of server resources on the acceptable flow rate. The results demonstrate that the proposed algorithm can achieve near-optimal performance and can significantly increase the acceptable flow rate and the service capacity compared to other algorithms under an energy cost budget.

Fixed and mobile network operators increasingly deploy managed content distribution networks (CDNs) with the objective of reducing the traffic on their transit links and to improve their customers' quality of experience. As network operator managed CDNs (nCDNs) become commonplace, operators will likely provide common interfaces to interconnect their nCDNs for mutual benefit, as they do with peering today. In this paper, we consider the problem of using distributed algorithms for computing a cache allocation for nCDNs. We show that if every network operator aims to minimize its cost and bilateral payments are not allowed, then it may be impossible to compute a cache allocation. For the case when bilateral payments are possible, we propose two distributed algorithms, the aggregate value compensation and the object value compensation algorithms, which differ in terms of the level of parallelism they allow and in terms of the amount of information exchanged between nCDNs. We prove that the algorithms converge, and we propose a scheme to ensure ex-post individual rationality. Simulations performed on a real autonomous system-level network topology and synthetic topologies show that the algorithms have geometric rate of convergence, and scale well with the graphs' density and the nCDN capacity.

Mobile collaborative community (MCC) is an emerging technology that allows multiple mobile nodes (MNs) to perform a resource intensive task, such as large content download, in a cooperative manner. In this paper, we introduce a proxy-based collaboration system for the MCC where a content proxy (CProxy) determines the amount of chunks and the sharing order scheduled to each MN, and the received chunks are shared among MNs via Wi-Fi Direct. We formulate a multi-objective optimization problem to minimize both the collaborative content download time and the energy consumption in an MCC, and propose a heuristic algorithm for solving the optimization problem. Extensive simulations are carried out to evaluate the effects of the number of MNs, the wireless bandwidth, the content size, and dynamic channel conditions on the content download time and the energy consumption. Our results demonstrate that the proposed algorithm can achieve near-optimal performance and significantly reduce the content download time and has an energy consumption comparable to that of other algorithms.

Motivated by emerging vision-based intelligent services, we consider the problem of rate adaptation for high-quality and low-delay visual information delivery over wireless networks using scalable video coding. Rate adaptation in this setting is inherently challenging due to the interplay between the variability of the wireless channels, the queuing at the network nodes, and the frame-based decoding and playback of the video content at the receiver at very short time scales. To address the problem, we propose a low-complexity model-based rate adaptation algorithm for scalable video streaming systems, building on a novel performance model based on stochastic network calculus. We validate the analytic model using extensive simulations. We show that it allows fast near-optimal rate adaptation for fixed transmission paths, as well as cross-layer optimized routing and video rate adaptation in mesh networks, with less than 10% quality degradation compared to the best achievable performance.

We consider the problem of performing distributed visual analysis for a video sequence in a visual sensor network that contains sensor nodes dedicated to processing. Visual analysis requires the detection and extraction of visual features from the images, and thus the time to complete the analysis depends on the number and on the spatial distribution of the features, both of which are unknown before performing the detection. In this paper, we formulate the minimization of the time needed to complete the distributed visual analysis for a video sequence subject to a mean average precision requirement as a stochastic optimization problem. We propose a solution based on two composite predictors that reconstruct randomly missing data, on quantile-based linear approximation of feature distribution and on time series analysis methods. The composite predictors allow us to compute an approximate optimal solution through linear programming. We use two surveillance video traces to evaluate the proposed algorithms, and show that prediction is essential for minimizing the completion time, even if the wireless channel conditions vary and introduce significant randomness. The results show that the last value predictor together with regular quantile-based distribution approximation provide a low complexity solution with very good performance.

A future content-centric Internet would likely consist of autonomous systems (ASes) just like today's Internet. It would thus be a network of interacting cache networks, each of them optimized for local performance. To understand the influence of interactions between autonomous cache networks, in this paper, we consider ASes that maintain peering agreements with each other for mutual benefit and engage in content-level peering to leverage each others' cache contents. We propose a model of the interaction between the caches managed by peering ASes. We address whether stable and efficient content-level peering can be implemented without explicit coordination between the neighboring ASes. We show that content-level peering leads to stable cache configurations, both with and without coordination. However, peering Internet Service Providers (ISPs) that coordinate to avoid simultaneous updates converge to a stable configuration more efficiently. Furthermore, if the content popularity estimates are inaccurate, content-level peering is likely to lead to cost efficient cache allocations. We validate our analytical results using simulations on the measured peering topology of more than 600 ASes.

Many Internet service providers (ISPs) have deployed peer-to-peer (P2P) caches in their networks in order to decrease costly inter-ISP traffic. A P2P cache stores parts of the most popular contents locally, and if possible serves the requests of local peers to decrease the inter-ISP traffic. Traditionally, P2P cache resource management focuses on managing the storage resource of the cache so as to maximize the inter-ISP traffic savings. In this paper, we show that when there are many overlays competing for the upload bandwidth of a P2P cache, then in order to maximize the inter-ISP traffic savings, the cache's upload bandwidth should be actively allocated among the overlays. We formulate the problem of P2P cache bandwidth allocation as a Markov decision process and propose three approximations to the optimal cache bandwidth allocation policy. We use extensive simulations and experiments to evaluate the performance of the proposed policies, and show that the bandwidth allocation policy that prioritizes swarms with a small ratio of local peers to all peers in the swarm can improve the inter-ISP traffic savings in BitTorrent-like P2P systems by up to 30%-60%.

Caching at the network edge is considered a promising solution for addressing the ever-increasing traffic demand of mobile devices. The problem of proactive content replication in hierarchical cache networks, which consist of both network edge and core network caches, is considered in this paper. This problem arises because network service providers wish to efficiently distribute content so that user-perceived performance is maximized. Nevertheless, current high-complexity replication algorithms are impractical due to the vast number of involved content items. Clustering algorithms inspired from machine learning can be leveraged to simplify content replication and reduce its complexity. Specifically, similar items could be clustered together, e.g., according to their popularity in space and time. Replication on a cluster-level is a problem of substantially smaller dimensionality, but it may result in suboptimal decisions compared to item-level replication. The factors that cause performance loss are identified and a clustering scheme that addresses the specific challenges of content replication is devised. Extensive numerical evaluations, based on realistic traffic data, demonstrate that for reasonable cluster sizes the impact on actual performance is negligible.

In-motion inductive charging, or dynamic charging, is an emerging technology that allows electric vehicles (EVs) to be charged while on the move. Accurate billing for dynamic EV charging requires secure communication between the EVs and the utility, and could potentially require the secure delivery of small messages from the EVs to the utility at a very high rate, which is infeasible with the currently available solutions. In this paper we propose Fast Authentication for Dynamic EV Charging (FADEC) designed to meet the communication needs of in-motion inductive EV charging. FADEC features fast signing and verification, low communication overhead, and fast hand-off authentication to support EV mobility. Our simulations show that compared with ECDSA mandated by 802.11p standard, FADEC reduces data delivery delay by up to 97%, increases the data delivery ratio by more than an order of magnitude and enables timely data delivery even in a resource constrained environment.

Internet service providers increasingly deploy internal CDNs with the objective of reducing the traffic on their transit links and to improve their customers' quality of experience. Once ISP managed CDNs (nCDNs) become commonplace, ISPs would likely provide common interfaces to interconnect their nCDNs for mutual benefit, as they do with peering today. In this paper we consider the problem of using distributed algorithms for computing a content allocation for nCDNs. We show that if every ISP aims to minimize its cost and bilateral payments are not allowed then it may be impossible to compute a content allocation. For the case of bilateral payments we propose two distributed algorithms, the aggregate value compensation (AC) and the object value compensation (OC) algorithms, which differ in terms of the level of parallelism they allow and in terms of the amount of information exchanged between nCDNs. We prove that the algorithms converge, and we propose a scheme to ensure ex-post individual rationality. Simulations performed on a real AS-level network topology and synthetic topologies show that the algorithms have geometric rate of convergence, and scale well with the graphs' density and the nCDN capacity.

Nodes in a sensor network are traditionally used for sensing and data forwarding. However, with the increase of their computational capability, they can be used for in-network data processing, leading to a potential increase of the quality of the networked applications as well as the network lifetime. Visual analysis in sensor networks is a prominent example where the processing power of the network nodes needs to be leveraged to meet the frame rate and the processing delay requirements of common visual analysis applications. The modeling of the end-to-end performance for such networks is, however, challenging, because in-network processing violates the flow conservation law, which is the basis for most queuing analysis. In this work we propose to solve this methodological challenge through appropriately scaling the arrival and the service processes, and we develop probabilistic performance bounds using stochastic network calculus. We use the developed model to determine the main performance bottlenecks of networked visual processing. Our numerical results show that an end-to-end delay of 2-3 frame length is obtained with violation probability in the order of 10-6. Simulation shows that the obtained bounds overestimates the end-to-end delay by no more than 10%.

Real-time visual analysis tasks, like tracking and recognition, require swift execution of computationally intensive algorithms. Enabling visual sensor networks to perform such tasks can be achieved by augmenting the sensor network with processing nodes and distributing the computational burden among several nodes, in a way that the cameras contend for the processing nodes while trying to minimize their completion times. In this paper, we formulate the problem of minimizing the completion time of all camera sensors as an optimization problem. We propose algorithms for fully distributed optimization, analyze the existence of equilibrium allocations, and evaluate their performance. Simulation results show that distributed optimization can provide good performance despite limited information availability at low computational complexity, but the predictable and stable performance is often not provided by the algorithm that provides lowest average completion time.

We study the statistical characteristics of SURF and BRISK interest points and descriptors, with the aim of supporting the design of distributed processing across sensor nodes in a resource -constrained visual sensor network (VSN). Our results show high variability in the density, the spatial distribution , and the octave layer distribution of the interest points. The high variability implies that balancing the processing load among the sensor nodes is a very challenging task, and obtaining a priori information is essential, e.g., through prediction . Our results show that if a priori information is available about the images, then Top- M interest point selection, limited , octave-based processing at the camera node, together with area-based interest point detection and extraction at the processing nodes, can balance the processing load and limit the transmission cost in the network . Complete interest point detection at the camera node with optimized descriptor extraction delegation to the processing nodes in turn can further decrease the transmission load and allow a better balance of the processing load among the network nodes.

Enabling visual sensor networks to perform visual analysis tasks in real-time is challenging due to the computational complexity of detecting and extracting visual features. A promising approach to address this challenge is to distribute the detection and the extraction of local features among the sensor nodes, in which case the time to complete the visual analysis of an image is a function of the number of features found and of the distribution of the features in the image. In this paper we formulate the minimization of the time needed to complete the distributed visual analysis for a video sequence subject to a mean average precision requirement as a stochastic optimization problem. We propose a solution based on two composite predictors that reconstruct randomly missing data, and use a quantile-based linear approximation of the feature distribution and time series analysis methods. The composite predictors allow us to compute an approximate optimal solution through linear programming. We use two surveillance videos to evaluate the proposed algorithms, and show that prediction is essential for controlling the completion time. The results show that the last value predictor together with regular quantile-based distribution approximation provide a low complexity solution with very good performance.

We consider controlling and balancing the processing load in a visual sensor network (VSN) used for detecting local features, such as BRISK. We formulate a prediction problem with random missing data, and propose two regression-based algorithms for data reconstruction. Numerical results illustrate the performance of the proposed algorithms, and show that backward regression combined with the last value predictor can be used for controlling and balancing the processing load in VSNs with good performance.

This demo showcases some of the results obtained by the GreenEyes project, whose main objective is to enable visual analysis on resource-constrained multimedia sensor networks. The demo features a multi-hop visual sensor network operated by BeagleBones Linux computers with IEEE 802.15.4 communication capabilities, and capable of recognizing and tracking objects according to two different visual paradigms. In the traditional compress-then-analyze (CTA) paradigm, JPEG compressed images are transmitted through the network from a camera node to a central controller, where the analysis takes place. In the alternative analyze-then-compress (ATC) paradigm, the camera node extracts and compresses local binary visual features from the acquired images (either locally or in a distributed fashion) and transmits them to the central controller, where they are used to perform object recognition/tracking. We show that, in a bandwidth constrained scenario, the latter paradigm allows to reach better results in terms of application frame rates, still ensuring excellent analysis performance.

Motivated by improved models for content workload prediction, in this paper we consider the problem of dynamic content allocation for a hybrid content delivery system that combines cloud-based storage with low cost dedicated servers that have limited storage and unmetered upload bandwidth. We formulate the problem of allocating contents to the dedicated storage as a finite horizon dynamic decision problem, and show that a discrete time decision problem is a good approximation for piecewise stationary workloads. We provide an exact solution to the discrete time decision problem in the form of a mixed integer linear programming problem, propose computationally feasible approximations, and give bounds on their approximation ratios. Finally, we evaluate the algorithms using synthetic and measured traces from a commercial music on-demand service and give insight into their performance as a function of the workload characteristics.

State estimation (SE) plays an essential role in the monitoring and supervision of power systems. In today's power systems, SE is typically done in a centralized or in a hierarchical way, but as power systems will be increasingly interconnected in the future smart grid, distributed SE will become an important alternative to centralized and hierarchical solutions. As the future smart grid may rely on distributed SE, it is essential to understand the potential vulnerabilities that distributed SE may have. In this paper, we show that an attacker that compromises the communication infrastructure of a single control center in an interconnected power system can successfully perform a denial-of-service attack against state-of-the-art distributed SE, and consequently, it can blind the system operators of every region. As a solution to mitigate such a denial-of-service attack, we propose a fully distributed algorithm for attack detection. Furthermore, we propose a fully distributed algorithm that identifies the most likely attack location based on the individual regions' beliefs about the attack location, isolates the identified region, and then reruns the distributed SE. We validate the proposed algorithms on the IEEE 118 bus benchmark power system.

A future content-centric Internet would likely consist of autonomous systems (ASes) just like today's Internet. It would thus be a network of interacting cache networks, each of them optimized for local performance. To understand the influence of interactions between autonomous cache networks, in this paper we consider ASes that maintain peering agreements with each other for mutual benefit, and engage in content-level peering to leverage each others' cache contents. We propose a model of the interaction and the coordination between the caches managed by peering ASes. We address whether stable and efficient content-level peering can be implemented without explicit coordination between the neighboring ASes or alternatively, whether the interaction needs to rely on explicit announcements of content reachability in order for the system to be stable. We show that content-level peering leads to stable cache configurations, both with and without coordination. If the ASes do coordinate, then coordination that avoids simultaneous updates by peering ISPs provides faster and more cost efficient convergence to a stable configuration. Furthermore, if the content popularity estimates are inaccurate, content-level peering is likely to lead to cost efficient cache allocations. We validate our analytical results using simulations on the measured peering topology of more than 600 ASes.

With BitTorrent, efficient peer upload utilization is achieved by splitting contents into many small pieces, each of which may be downloaded from different peers within the same swarm. Unfortunately, piece and bandwidth availability may cause the file-sharing efficiency to degrade in small swarms with few participating peers. Using extensive measurements, we identified hundreds of thousands of torrents with several small swarms for which reallocating peers among swarms and/or modifying the peer behavior could significantly improve the system performance. Motivated by this observation, we propose a centralized and a distributed protocol for dynamic swarm management. The centralized protocol (CSM) manages the swarms of peers at minimal tracker overhead. The distributed protocol (DSM) manages the swarms of peers while ensuring load fairness among the trackers. Both protocols achieve their performance improvements by identifying and merging small swarms and allow load sharing for large torrents. Our evaluations are based on measurement data collected during eight days from over 700 trackers worldwide, which collectively maintain state information about 2.8 million unique torrents. We find that CSM and DSM can achieve most of the performance gains of dynamic swarm management. These gains are estimated to be up to 40% on average for small torrents.

Many Internet service providers (ISPs) have deployed peer-to-peer (P2P) caches in their networks in order to decrease costly inter-ISP traffic. A P2P cache stores parts of the most popular contents locally, and if possible serves the requests of local peers to decrease the inter-ISP traffic. Traditionally, P2P cache resource management focuses on managing the storage resource of the cache so as to maximize the inter-ISP traffic savings. In this paper we show that when there are many overlays competing for the upload bandwidth of a P2P cache then in order to maximize the inter-ISP traffic savings the cache's upload bandwidth should be actively allocated among the overlays. We formulate the problem of P2P cache bandwidth allocation as a Markov decision process, and describe two approximations to the optimal cache bandwidth allocation policy. Based on the insights obtained from the approximate policies we propose SRP, a priority-based allocation policy for BitTorrent-like P2P systems. We use extensive simulations to evaluate the performance of the proposed policies, and show that cache bandwidth allocation can improve the inter-ISP traffic savings by up to 30 to 60 percent. We validate the results via BitTorrent experiments on Planet-lab.

With the proliferation of cloud services, cloud-based systems can become a cost-effective means of on-demand content delivery. In order to make best use of the available cloud bandwidth and storage resources, content distributors need to have a good understanding of the tradeoffs between various system design choices. In this work we consider a peer-assisted content delivery system that aims to provide guaranteed average download rate to its customers. We show that bandwidth demand peaks for contents with moderate popularity, and identify these contents as candidates for cloud-based service. We then consider dynamic content bundling (inflation) and cross-swarm seeding, which were recently proposed to improve download performance, and evaluate their impact on the optimal choice of cloud service use. We find that much of the benefits from peer seeding can be achieved with careful torrent inflation, and that hybrid policies that combine bundling and peer seeding often reduce the delivery costs by 20% relative to only using seeding. Furthermore, all these peer-assisted policies reduce the number of files that would need to be pushed to the cloud. Finally, we show that careful system design is needed if locality is an important criterion when choosing cloud-based service provisioning.

As a model of distributed resource allocation in networked systems, we consider resource allocation games played over a influence graph. The influence graph models limited interaction between the players due to, e.g., the network topology: the payoff that an allocated resource yields to a player depends only on the resources allocated by her neighbors on the graph. We prove that pure strategy Nash equilibria (NE) always exist in graphical resource allocation games and we provide a linear time algorithm to compute equilibria. We show that these games do not admit a potential function: if there are closed paths in the influence graph then there can be best reply cycles. Nevertheless, we show that from any initial allocation of a resource allocation game it is possible to reach a NE by playing best replies and we provide a bound on the maximal number of update steps required. Furthermore we give sufficient conditions in terms of the influence graph topology and the utility structure under which best reply cycles do not exist. Finally we propose an efficient distributed algorithm to reach an equilibrium over an arbitrary graph and we illustrate its performance on different random graph topologies.

Critical power system applications like contingency analysis and optimal power flow calculation rely on the power system state estimator. Hence the security of the state estimator is essential for the proper operation of the power system. In the future more applications are expected to rely on it, so that its importance will increase. Based on realistic models of the communication infrastructure used to deliver measurement data from the substations to the state estimator, in this paper we investigate the vulnerability of the power system state estimator to attacks performed against the communication infrastructure. We define security metrics that quantify the importance of individual substations and the cost of attacking individual measurements. We propose approximations of these metrics, that are based on the communication network topology only, and we compare them to the exact metrics. We provide efficient algorithms to calculate the security metrics. We use the metrics to show how various network layer and application layer mitigation strategies, like single and multi-path routing and data authentication, can be used to decrease the vulnerability of the state estimator. We illustrate the efficiency of the algorithms on the IEEE 118 and 300 bus benchmark power systems.

Peer-to-peer file-sharing systems are responsible for a significant share of the traffic between Internet service providers (ISPs) in the Internet. In order to decrease their peer-to-peer-related transit traffic costs, many ISPs have deployed caches for peer-to-peer traffic in recent years. We consider how the different types of peer-to-peer caches—caches already available on the market and caches expected to become available in the future—can possibly affect the amount of inter-ISP traffic. We develop a fluid model that captures the effects of the caches on the system dynamics of peer-to-peer networks and show that caches can have adverse effects on the system dynamics depending on the system parameters. We combine the fluid model with a simple model of inter-ISP traffic and show that the impact of caches cannot be accurately assessed without considering the effects of the caches on the system dynamics. We identify scenarios when caching actually leads to increased transit traffic. Motivated by our findings, we propose a proximity-aware peer-selection mechanism that avoids the increase of the transit traffic and improves the cache efficiency. We support the analytical results by extensive simulations and experiments with real BitTorrent clients.

Motivated by applications in industrial communication networks, in this paper we consider the trade-off between relationship anonymity and communication overhead in anonymity networks. We consider two anonymity networks; Crowds that provides unbounded communication delay and Minstrels, proposed in this paper, that provides bounded communication delay. While Crowds hides the sender's identity only, Minstrels aims to hide the receiver's identity as well. However, to achieve bounded message delay it has to expose the sender's identity to a greater extent than Crowds. We derive exact and approximate analytical expressions for the relationship anonymity for these systems. While Minstrels achieves close to optimal anonymity under certain conditions, our results show that, contrary to expectations, increased overhead does not always improve anonymity.

Tracker-based peer-discovery is used in most commercial peer-to-peer content distribution systems, as it provides performance benefits compared to distributed solutions, and facilitates the control and monitoring of the overlay. But a tracker is a central point of failure, and its deployment and maintenance incur costs; hence an important question is how high tracker availability can be achieved at low cost. We investigate highly available, low overhead peer discovery, using independent trackers and a simple gossip protocol. This work is a step towards understanding the trade-off between the overhead and the achievable peer connectivity in highly available distributed overlay-management systems for peer-to-peer content distribution. We propose two protocols that connect peers in different swarms efficiently with a constant, but tunable, overhead. The two protocols, Random Peer Migration (RPM) and Random Multi-Tracking (RMT), employ a small fraction of peers in a torrent to virtually increase the size of swarms. We develop analytical models of the protocols based on renewal theory, and validate the models using both extensive simulations and controlled experiments. We illustrate the potential value of the protocols using large-scale measurement data that contains hundreds of thousands of public torrents with several small swarms, with limited peer connectivity. We estimate the achievable gains to be up to 40% on average for small torrents.

We consider whether cooperative caching may reduce the transit traffic costs of Internet service providers (ISPs) due to peer-to-peer (P2P) content distribution systems. We formulate two game-theoretic models for cooperative caching, one in which ISPs follow their selfish interests, and one in which they act altruistically. We show the existence of pure strategy Nash equilibria for both games, and evaluate the gains of cooperation on various network topologies, among them the AS level map of Northern Europe, using measured traces of P2P content popularity. We find that cooperation can lead to significant improvements of the cache efficiency with little communication overhead even if ISPs follow their selfish interests.

We consider how relatively simple extensions of popular channel-aware schedulers can be used to multicast scalable video streams in high speed radio access networks. To support the evaluation, we first describe a model of the channel distortion of scalable video coding and validate it using eight commonly used test sequences. We use the distortion model in a detailed simulation setup to compare the performance of six schedulers, among them the Max-Sum and Max-Prod schedulers, which aim to maximize the sum and the product of streaming utilities, respectively. We investigate how the traffic load, user mobility, layering structure, and users' aversion of fluctuating distortion influence the streaming performance. Our results show that the Max-Sum scheduler performs better than other considered schemes in almost all scenarios. With the Max-Sum scheduler, the gain of scalable video coding compared to non-scalable coding is substantial, even when users do not tolerate frequent changes in video quality.

Peer-to-peer file-sharing systems are responsible for a significant share of the traffic between Internet service providers (ISPs) in the Internet. In order to decrease their peer-to-peer related transit traffic costs, many ISPs have deployed caches for peer-to-peer traffic in recent years. We consider how the different types of peer-to-peer caches - caches already available on the market and caches expected to become available in the future - can possibly affect the amount of inter-ISP traffic. We develop a fluid model that captures the effects of the caches on the system dynamics of peer-to-peer networks, and show that caches can have adverse effects on the system dynamics depending on the system parameters. We combine the fluid model with a simple model of inter-ISP traffic and show that the impact of caches cannot be accurately assessed without considering the effects of the caches on the system dynamics. We identify scenarios when caching actually leads to increased transit traffic. Our analytical results are supported by extensive simulations and experiments with real BitTorrent clients.

A large number of peer-to-peer streaming systems have been proposed and deployed in recent years. Yet, there is no clear understanding of how these systems scale and how multipath and multihop transmission, properties of all recent systems, affect the quality experienced by the peers. In this paper, we present an analytical study that considers the relationship between delay and loss for general overlays: we study the trade-off between the playback delay and the probability of missing a packet and we derive bounds on the scalability of the systems. We present an exact model of push-based overlays and show that the bounds hold under diverse conditions: in the presence of errors, under node churn, and when using forward error correction and various retransmission schemes.

In this paper we propose and analyze a generalized multiple-tree-based overlay architecture for peer-to-peer live streaming that employs multipath transmission and forward error correction. We give mathematical models to describe the stability properties of the overlay and evaluate the error recovery in the presence of node dynamics and packet losses. We show how the stability of the overlay improves with the proper allocation of the outgoing bandwidths of the peers among the trees without compromising its error correcting capability.

Applications that require low loss probabilities in today's Internet have to employ some end-to-end error-recovery mechanism. For interactive applications with strict delay constraints, the delay introduced by the applied schemes has to be low as well. In this paper we compare two schemes proposed for error recovery for real-time audio applications: media-dependent forward error correction (MD-FEC) and multiple description coding (MDC). We conclude that MDC always performs better than MD-FEC, and that the stationary loss probability plays a key role in the choice of the optimal parameters for these schemes. Combining the analytical results with the loss characteristics of measured traces of VoIP calls we conclude that in the current Internet these schemes give considerable gains for streams with a high code rate only, and for these streams MDC can decrease the average distortion significantly better than MD-FEC.

In this paper we propose an analytical model of a resilient end-node multicast streaming architecture based on multiple minimum-depth-trees that employs path diversity and forward error correction for improved resilience to node churns and packet losses. We study the performance of the architecture in the presence of packet losses and dynamic node behavior. We show that for a given redundancy the probability that an arbitrary node possesses a packet is high as long as the loss probability in the network is below a certain threshold. After reaching the threshold the packet possession probability suddenly drops; the rate decrease gets faster as the number of nodes in the overlay grows. The value of the threshold depends on the ratio of redundancy and on the number of the distribution trees. We study the overlay structure in the presence of node dynamics and conclude that stability can be achieved only if the root node serves a large number of nodes simultaneously