Graph Neural Networks (GNNs) have gained significant attention for multivariate time series analysis in recent years. However, applying them to real-world networking data introduces several key challenges, particularly in designing meaningful and effective graph structures. In this paper, we propose a novel method for constructing initial graphs tailored for time series anomaly detection in complex network environments. Our approach, called COSI (COrrelation SImilarity), leverages two fundamental properties of real-world data: feature name semantics and statistical correlation. By combining natural language processing (NLP) with correlation analysis, COSI produces graph structures that significantly enhance GNN model performance for anomaly detection tasks, outperforming conventional graph construction methods in almost all evaluation scores across all datasets. We extensively evaluate COSI on three datasets, including two real-world networking datasets and one widely used benchmark dataset, and we open source the implementation to encourage reproducibility, further research, and practical adoption by the community.

Applying Federated Learning (FL) to real-world network environments, such as telecommunication networks, presents significant challenges. In this paper, we investigate the intrinsic problem of non-IID data training across distributed innetwork FL clients. More specifically, we focus on the feature distribution discrepancy in the use of in-network FL for infrastructure monitoring. While previous works have made notable progress in the design of aggregation functions that compensate strong polarization in data distributions, limited attention was directed toward data load-balancing from sources to processing nodes, which leverages the role and characteristics of the data itself. Our goal is to address feature heterogeneity in in-network federated learning by piloting how data is forwarded in the network on the way to its consumers. To this end, we design and evaluate several scenarios for dynamic data load balancing between FL clients within the network topology, subject to latency constraints.

The Open Radio Access Network (O-RAN) paradigm has attracted considerable attention from both academic researchers and industry stakeholders, owing to its capacity to significantly enhance the adaptability and flexibility of cellular network architectures. However, the disaggregation of Generation NodeB (gNB) in O-RAN introduces increased system complexity and potential variability in performance metrics, necessitating empirical evaluation to determine whether its performance can rival or exceed that of conventional monolithic designs. In this study, we developed a standalone (SA) testbed to quantitatively assess the downlink (DL) transmission performance of two O-RAN architectures, implemented on two widely adopted open-source platforms: srsRAN and OAI. To ensure a scientifically robust and comprehensive analysis, we meticulously selected key performance metrics, implemented measures to provide solid statistical results and eliminate residual effects between test iterations. Test results show that the split enhances network flexibility by offloading certain processing tasks to the Distributed Unit (DU), thereby reducing the overall hardware usage burden and potentially improving the computational efficiency and resource allocation in future RAN networks.

In this paper, we model carrier shutdown in multi-carrier mobile networks as a deep reinforcement learning problem. Our model takes energy-saving actions by turning off carriers and reallocating their users while in addition to maintaining connectivity guarantees a novel user experience metric. Leveraging real and recent datasets, we train and evaluate our model over realistic network scenarios. Our results show more than 15% energy saving with the fulfillment of the user experience constraints, outperforming currently deployed solutions and researched approaches in the literature by almost 50%. More interestingly our approach exhibits generalization properties, a very promising characteristic for its adoption in real mobile networks deployment.

In this paper, we address the challenge of anomaly detection in multivariate time series data related to environments requiring stringent guarantees, notably in 5G and beyond 5G systems promising high reliability and quality. Traditional approaches to anomaly detection, including supervised, unsuper-vised, and self-supervised methods often struggle with the diver-sity and unpredictability of real-world anomalies. To overcome these limitations, we propose a novel dual forecasting model approach, DREAM (Dual foREcAsting Model Anomaly Detection), which leverages both normal and unlabeled data to enhance detection accuracy. Our approach involves training two distinct models: one on normal behavior and the other on mixed behavior, and then comparing their outputs to identify anomalies. We also introduce new evaluation method, addressing the shortcomings of traditional point-wise evaluation. Our experiments, with multiple networked system datasets, demonstrate that DREAM outper-forms traditional forecasting-based approaches, including when used in a hybrid manner with traditional forecasting algorithms.

The rise of xG networks has brought unprecedented capabilities in wireless communication, but the complexity of 5G and beyond 5G networks introduces challenges in ensuring reliability and performance. In this article, we propose a new approach to network anomaly detection by leveraging Graph Neural Networks (GNNs) and graph structure learning. GNNs are well-suited for capturing relationships in graph-structured data, making them an effective tool for detecting complex patterns in network behavior. We introduce a novel graph structure extracting feature semantics and demonstrate the effectiveness of GNNs in network anomaly detection. We show we can improve the accuracy up to 4% thanks to the proposed graph structure. The source code is available at https://github.com/Killian-cressant/graph4_anomaly_detection.

To address rising Quality of Service (QoS) requirements in beyond 5G systems, this paper proposes a data-centric approach for user plane QoS control in beyond 5G systems. Building upon the PSBA-XG platform, previously designed for control plane optimization, we extend its capabilities to monitor and manage the User Plane Function (UPF) performance. By aggregating standard-compliant usage reports using a publish/subscribe architecture, our solution enables rapid detection of QoS degradation and reconfiguration of user sessions. Experimental results on an open-source 5G testbed demonstrate that our approach allows for more flexible QoS management, notably rerouting user traffic to edge resources in case of congestion, than a plain 5G system, without introducing additional complexity. These results highlight the potential of data-centric architectures to enhance QoS in future mobile network systems.

This paper introduces the design of a data pipeline system (DPS) integrated with artificial intelligence (AIF) functions to support continuous AI learning and operations for network automation in 5G/6G systems. We design the DPS as a chain of functions, namely ingress and egress Network Data Broker Function (iNDBF and eNDBF) and Network Data Preprocessing Function (NDPPF), to support in-network learning operations. To take into account the distributed nature of the network architecture of 5G systems and beyond, we conceive the DPS to be integrated seamlessly with a distributed learning frameworks such as the federated learning (FL). We performed a realistic evaluation, employing a real dataset from a national mobile operator to simulate the network architecture. Additionally, a FL framework for anomaly detection is integrated with the DPS to assess the effectiveness of our proposal. Evaluation results show that delays in end-to-end data transmission and preprocessing to the AIF locations can cause distributed learning AIFs to work with stale data. The results also highlight how the DPS can counterbalance these delays leading to desynchronisation of the distributed learning process, bringing to AIFs with higher accuracy.

In this paper, we formulate the traffic-aware mobile nodes sleeping with traffic offloading as a Markov Decision Process (MDP) and solve it using Deep Reinforcement Learning (DRL). Our model characterizes jointly the energy saving actions due to base stations entering in sleep mode as well offloading options to neighboring nodes of the turned off gNodeB. To solve this problem, the Proximal Policy Optimization (PPO) integrated with action masking is leveraged. Our validation results, when training the model with open source datasets, show a potential of reducing up to 16% of the network energy consumption without negatively affecting traffic coverage.

Network slicing allows leveraging virtualization techniques for the creation of multiple, logically-isolated network instances over a shared infrastructure. In general, it is composed of a set of unique network functions with a specific physical capacity request. In order to improve network and service robustness, ETSI has introduced the concept of Network Function Set where network functions are replicated and deployed in different physical nodes. In this paper, we integrate this concept of the network function set to implement load-balancing and efficient resource distribution in 5G networks by leveraging on an existing network slice design formulation without the network function set. This helps relieve the burden on the nodes and links and prevent QoS degradation, in particular during failures. We describe how the baseline approach is impacted for the placement of network functions. We then show that our approach improves load balancing and latency with respect to the baseline solution.

Botnets pose a significant threat in network security, exacerbated by the massive adoption of vulnerable Internet-of-Things (IoT) devices. In response to that, great research effort has taken place to propose intrusion detection solutions to the botnet menace. As most techniques focus on either packet or flow granularity, port-based analysis can help detecting newly developed botnets, especially during their early propagation phase. In this paper, we introduce a line rate distributed anomaly detection system that employs NetFPGA Smart-Network Interface Cards (SmartNIC) as programmable switches. Per-port feature extraction modules are deployed directly on the data plane, enabling a centralized controller to periodically retrieve collected metrics, and feed them to a botnet detection algorithm we refine from the state of the art. We evaluate our system using real world traces spanning several months from 2016 and 2023. We show how our solutions allow keeping low the number of anomalies detected, retaining only the most relevant ones, thanks to the distributed monitoring approach that helps discriminating systemic changes from local phenomena. Furthermore, we provide an analysis of the most significant alerts, accounting for the limited ground-truth on the dataset.

In this paper, we present ELoRa, an emulation tool that generates Long Range Wide Area Network (LoRaWAN) traffic for an arbitrary number of LoRa devices and in an end-to-end virtualized LoRaWAN setting. Using ns-3, we improve an existing radio access network simulator to produce traffic compatible with ChirpStack, an open-source, cloud-native, LoRaWAN network functions stack. Our tool can be used to create realistic traffic and anomalies in order to test orchestration techniques on a real, distributed infrastructure. Moreover, the LoRaWAN core network functions (bridges, network server) are agnostic to the simulation of the radio access and can change parameters of simulated devices using native LoRaWAN protocol primitives, therefore enabling live-testing of resource allocation techniques to manage the radio access network. Multiple ELoRa instances can be connected to the same LoRaWAN core, each instance being able to support 50000 devices and 7 gateways.

Edge intelligence combined with federated learning is considered as a way to distributed learning and inference tasks in a scalable way, by analyzing data close to where it is generated, unlike traditional cloud computing where data is offloaded to remote servers. In this paper, we address the placement of Artificial Intelligence Functions (AIF) making use of federated learning and hardware acceleration. We model the behavior of federated learning and related inference point to guide the placement decision, taking into consideration the specific constraint and the empirical behavior of a virtualized infrastructure anomaly detection use-case. Besides hardware acceleration, we consider the specific training time trend when distributing training over a network, by using empirical piece-wise linear distributions. We model the placement problem as a MILP and we propose a variant of the problem. Simulation results show the impact that hardware acceleration can have in the decision of the number of AIF to enable, while dividing by a relevant factor the distributed training time. We also show how our approach exacerbates the importance of monitoring an end-to-end learning system delay budget composed of link propagation delay and distributed training time in the location of AIFs.

Motivated by the rapid deployment of applications based on connected objects, we propose in this paper an approach to differentiating Packet Delivery Ratios (PDRs) in LoRa Wide Area Networks (LoRaWAN). This type of network is simple to deploy and operate at the expense of loose commitments in terms of quality. To overcome this shortcoming, we propose an access control method for isolating clusters of devices and meeting differentiated PDR targets. Results show that our method outperforms known one in terms of PDR via improved parameter allocation and achieves high level of intra-cluster fairness, at the expense however of decreasing the maximum cell range.

Service quality differentiation is gaining popularity in IoT networks, notably in LoRaWAN, with the rapid widespread of applications on connected devices. There is clearly a business demand for quality in IoT in the context of smart cities and network operators are urged by application designer to offer quality differentiation. However, those types of networks have been designed on the basis of a best effort service model. In particular, Packet Delivery Ratio (PDR) can dramatically decrease in dense scenarios. In this paper, we propose and evaluate traffic control and channel assignment solutions for PDR differentiation in dense deployments. Several performance criteria are defined in order to analyze the gain achieved by a network operator as well as end users. Numerical results show that both players can benefit from quality differentiation with ad-hoc pricing. This proves to be effective if penalizing low requirement devices, as they can create a bottleneck in the system. Namely, we show that in high density settings we can reach a 20% better PDR with one of the proposed policies, improving mean device servicing rate by 10% and the operator gain by 7.5%.

In this work, we provide a Machine Learning framework for augmenting the Differentiated Services (DiffServ) protocol with fine-grained dynamic traffic classification. The framework is called L-DiffServ. It is composed of two classification algorithms able to detect the QoS classes of incoming packets only looking at three packet header fields; the first algorithm, referred to as Inter-L-DiffServ, is a semi-supervised classification procedure able to replicate DiffServ classification; the second one, referred to as Intra-L-DiffServ, is an unsupervised algorithm for intra-class classification, useful for classes taking large portions of the overall traffic. We apply the latter to the low priority best-effort class. The performance evaluation shows that our solution is able to dynamically classify packets and to detect new QoS sub-classes hence adapting to traffic aggregate characteristics. We also show that network resource management can be improved exploiting the new generated QoS sub-classes: two active queue management algorithms based on WRED and CHOKe show a reduction of the number of sessions affected by packet losses up to 40% with respect to the legacy DiffServ procedure.

We model the network slice provisioning as an optimization problem including novel mapping and provisioning requirements rising with new radio and core function placement policies. We propose an open-access framework based on an MILP formulation that encompasses flexible functional splitting, with possibly different splitting for different slices and slice subnets, while taking into account different network sharing policies from 5G specifications. We also consider novel mapping and continuity constraints specific to the 5G architectures and beyond. We show by numerical simulations the impact of taking into full and partial consideration these peculiar novel technical constraints.

Network softwarization technologies challenge legacy fault management systems. Coordination and dependency among different novel software components for orchestration, switching, virtual machine and container management creates novel monitoring points, besides novel sources of faults, bugs and vulnerabilities. To cope with the high heterogeneity and granularity of software components, we propose a modular network AI framework to detect anomalies, toward closed-loop automation. We design an AI-empowered anomaly detection framework able to assess the running state and the state deviations of a softwarized infrastructure, monitored through different features grouped depending on their layer and connect-compute stack component. Our framework learns the nominal working conditions of the infrastructure, with respect to which anomalies are detected, and characterized tracing back the layer and component root cause; it includes a network state assessment technique that leverages anomalies characterization through their most visible symptoms. We implement and validate the proposed framework through experimental tests on a containerized IP Multimedia Subsystem platform.

Multi-access Edge Computing (MEC) technology has emerged to overcome traditional cloud computing limitations, challenged by the new 5G services with heavy and heterogeneous requirements on both latency and bandwidth. In this work, we tackle the problem of clustering access points in MEC environments, introducing a set of clustering models to be deployed at the pre-provisioning phase. We go through extensive simulations on real-world traffic demands to evaluate the performance of the proposed solutions. In addition, we show how MEC hosts capacity violation can be decreased when integrating access points clustering into the orchestration model, by investigating on solution accuracy when applied on held-out users traffic demands. The obtained results show that our approach outperforms two state-of-the-art algorithms, reducing both memory usage and execution time, by 46% and 50%, respectively, in comparison to a baseline algorithm. It surpasses the two methods in gaining control over MEC hosts capacity usage for different maximum achieved occupancy levels on MEC hosts.

Novel network infrastructures require the distribution of computing and network resource control to meet stringent requirements in terms of latency, reliability and bitrate. 5G systems bring a key novelty in systems design that it the ‘network slice’as a new resource provisioning entity. A network slice is meant to serve end-to-end services as a composition of different network and system resources as radio, link, storage and computing resources. Conventionally, each resource is managed by a distinct decision-maker, platform, provider, orchestrator or controller. Naturally, centralized slice orchestration approaches are proposed in the literature, where a multi-domain orchestrator allocates the resources, for instance using a multi-resource allocation rule. Nonetheless, while simplifying the algorithmic approach, centralization can come at the expense of scalability and performance. In this article, we propose new ways to distribute the slice multi-resource resource allocation problem, using cascade and parallel resource allocations that are functionally compatible with novel software platforms. We also show how to adapt the proposed algorithms to make them able to guarantee service level agreements on the minimum resource needed, and to take into account deadline priority policy scheduling. We provide an exhaustive analysis of the advantages and disadvantages of the different approaches, including a numerical analysis for a realistic setting.

In this paper we propose a game theoretical framework in order to formally characterize the robustness of blockchains systems in terms of resilience to rational deviations and immunity to Byzantine behaviors. Our framework includes necessary and sufficient conditions for checking the immunity and resilience of games and an original technique for composing games that preserves the robustness of individual games. We prove the practical interest of our formal framework by characterizing the robustness of various blockchain protocols: Bitcoin (the most popular permissionless blockchain), Tendermint (the first permissioned blockchain used by the practitioners), Lightning Network, a side-chain protocol and a cross-chain swap protocol. For each one of the studied protocols we identify upper and lower bounds with respect to their resilience and immunity (expressed as no worse payoff than the initial state) face to rational and Byzantine behaviors.

In this work, we model the network slice provisioning as an optimization problem including novel mapping and provisioning requirements rising with new 5G radio and core function placement policies. We propose an MILP-based formulation that joins different functional splitting strategies with different network function sharing policies and novel mapping continuity constraints from 5G specifications. We show by numerical simulations the impact of taking into full and partial consideration these peculiar sets of novel technical constraints.

Malicious jamming launched by smart jammer, which attacks legitimate transmissions has been regarded as one of the critical security challenges in wireless communications. Thus, this paper exploits intelligent reflecting surface (IRS) to enhance anti-jamming communication performance and mitigate jamming interference by adjusting the surface reflecting elements at the IRS. Aiming to enhance the communication performance against smart jammer, an optimization problem for jointly optimizing power allocation at the base station (BS) and reflecting beamforming at the IRS is formulated. As the jamming model and jamming behavior are dynamic and unknown, a win or learn fast policy hill-climbing (WoLFCPHC) learning approach is proposed to jointly optimize the anti-jamming power allocation and reflecting beamforming strategy without the knowledge of the jamming model. Simulation results demonstrate that the proposed anti-jamming based-learning approach can efficiently improve both the the IRS-assisted system rate and transmission protection level compared with existing solutions.

In this paper we address the distributed cross-chain swap problem in the blockchain context where multiple agents exchange assets across multiple blockchain systems (e.g. trading Bitcoins for Litecoins or Ethers). We present a mathematical framework allowing to characterize blockchain swap protocols as the combination of a publishing and a commitment phase, where contracts are respectively published and then committed. We characterize the equilibria of existing cross-chain swap protocols (i.e., blockchain swap protocols exchanging assets among different blockchains). More precisely, we prove that following a swap protocol characterized by concurrent publishing of exchange contracts and snap (immediate) assets transfers is a Nash equilibrium. Furthermore, we prove that for protocols with a sequential publishing and commitment of the assets transfers, following the prescribed protocol is a sub-game perfect equilibrium.

Quality of Service (QoS) management in IP networks today relies on static configuration of classes of service definitions and related forwarding priorities. Packets are actually classified according to the DiffServ architecture based on the RFC 4594, typically thanks to static configuration or filters matching packet features, at network access equipment. In this paper, we propose a dynamic classification procedure, referred to as Learning-powered DiffServ (L-DiffServ), able to detect the distinctive characteristics of traffic and to dynamically assign service classes to IP packets. The idea is to apply semi-unsupervised Machine Learning techniques, such as Linear Discriminant Analysis (LDA) and K-Means, with a proper customization to take into account the issues related to packet-level analysis, i.e. unbalanced distribution of traffic among classes and selection of proper IP header related features. The performance evaluation highlights that L-DiffServ is able to change dynamically the classification outcome, providing an higher number of classes than DiffServ. This last result represents the first step toward a more granular differentiation of IP traffic.

Efficient bot detection is a crucial security matter and has been widely explored in the past years. Recent approaches supplant flow-based detection techniques and exploit graph-based features, incurring however in scalability issues in terms of time and space complexity. Bots exhibit specific communication patterns: they use particular protocols, contact specific domains, hence can be identified by analyzing their communication with the outside. To simplify the communication graph, we look at frequency distributions of protocol attributes capturing the specificity of botnets behaviour. In this paper, we propose a bot detection technique named BotFP, for BotFinger-Printing, which acts by (i) characterizing hosts behaviour with at-tribute frequency distribution signatures, (ii) learning behaviour of benign hosts and bots through a clustering technique, and (iii) classifying new hosts based on distances to labelled clusters. We validate our solution on the CTU-13 dataset, which contains 13 scenarios of bot infections, connecting to a Command-and-Control (C&C) channel and launching malicious actions such as port scanning or Denial-of-Service (DDoS) attacks. Our approach applies to various bot activities and network topologies. The approach is lightweight, can handle large amounts of data, and shows better accuracy than state-of-the-art techniques.

The 5G infrastructure brings a key novelty in networked systems design that is a new resource provisioning entity, the so-called "network slice". A network slice is meant to serve end-to-end services as a composition of different network and system resources as the radio, the link and a variety of computing resources (CPU, RAM, storage), generally each managed by a distinct decision-maker (platform, provider, orchestrator or controller). Naturally, centralized slice orchestration approaches have been proposed, where a multi-domain orchestrator allocates the resources, using a multi-resource allocation rule. Nonetheless, while simplifying the algorithmic approach, centralization can come at the expense of scalability and performance. In this paper, we propose new ways to decentralize the slice resource allocation problem, using cascade or parallel resource allocations. We provide an exhaustive analysis of the advantages and disadvantages of the different approaches together with a numerical analysis in a realistic environment.

Efficient bot detection is a crucial security matter and widely explored in the past years. Recent approaches supplant flow-based detection techniques and exploit graph-based features, incurring however in scalability issues, with high time and space complexity. Bots exhibit specific communication patterns: they use particular protocols, contact specific domains, hence can be identified by analyzing their communication with the outside. A way we follow to simplify the communication graph and avoid scalability issues is looking at frequency distributions of protocol attributes capturing the specificity of botnets behaviour. We propose a bot detection technique named BotFP, for BotFingerPrinting, which acts by (i) characterizing hosts behaviour with attribute frequency distribution signatures, (ii) learning benign hosts and bots behaviours through either clustering or supervised Machine Learning (ML), and (iii) classifying new hosts either as bots or benign ones, using distances to labelled clusters or relying on a ML algorithm. We validate BotFP on the CTU-13 dataset, which contains 13 scenarios of bot infections, connecting to a Command-and-Control (C&C) channel and launching malicious actions such as port scanning or Denial-of-Service (DDoS) attacks. Compared to state-of-the-art techniques, we show that BotFP is more lightweight, can handle large amounts of data, and shows better accuracy.

Among the novelties introduced by 5G networks, the formalization of the `network slice' as a resource allocation unit is an important one. In legacy networks, resources such as link bandwidth, spectrum, computing capacity are allocated independently of each other. In 5G environments, a network slice is meant to directly serve end-to-end services, or verticals: behind a network slice demand, a tenant expresses the need to access a precise service type, under a fully qualified set of computing and network requirements. The resource allocation decision encompasses, therefore, a combination of different resources. In this paper, we address the problem of fairly sharing multiple resources between slices, in the critical situation in which the network does not have enough resources to fully satisfy slice demands. We model the problem as a multi-resource allocation problem, proposing a versatile optimization framework based on the Ordered Weighted Average (OWA) operator, that takes into account different fairness approaches. We show how, adapting the OWA utility function, our framework can generalize classical single-resource allocation methods, existing multi-resource allocation solutions at the state of the art, and implement novel multi-resource allocation solutions. We compare analytically and by extensive simulations the different methods in terms of fairness and system efficiency.

With an increasing demand for flexible management in software-defined networks (SDNs), it becomes critical to minimize the network policy update time. Although major SDN controllers are now optimized for rapid network update at the control plane, there is still room for data plane optimization in terms of update time, when using TCAM-based physical SDN commodity-off-the-shelf switches. A slow update directly affects network performance and creates bottlenecks. To minimize the flow entry update time, a dependency graph, a kind of directed acyclic graph (DAG), can be used for the access management of flow entries at the switch. Thanks to the DAG, unnecessary entry movements, which are the main factor slowing down flow entry updates, can be avoided. However, existing algorithms show limitations when updates become very frequent. We propose a new flow entry update algorithm, called FastRule, that exploits a greedy strategy with an efficient data structure to accelerate flow entry update with a DAG approach. Moreover, we also adjust our algorithm for other flow table layouts to make it scalable. We elaborate on the correctness of FastRule and test our algorithm using a hardware switch. Compared with existing algorithms, the evaluation shows that our algorithm is about 100x faster than state-of-the-art solutions with a flow table of 1k size.

With an increasing demand for resilience in software-defined networks (SDN), it becomes critical to minimize service recovery delay upon route failures. Fast reroute (FRR) mechanisms are widely used in IP and MPLS networks by computing the recovery path before a failure occurs. The centralized control plane in SDN can potentially enhance path computation, so that FRR path computation can better scale in SDN than in traditional networks. However, the traditional FRR path computation algorithms could lead to a poor performance in large-scale SDN. The problem can become more severe for a highly dynamic network, which often sees dozens of failures or configuration changes in any single day. We propose a new algorithm that exploits pruned searching to quickly compute recovery paths for all-pair switches/hosts upon a link failure. For applications requiring stringent path robustness levels, we also extend this algorithm to quickly find the shortest guaranteed-cost path, which ensures that the recovery path used upon on-path link failures has the minimum cost. Compared with traditional solutions, our evaluations show that our algorithm is about 8 ~ 81 times faster than the practical implementation, 1.93 ~ 3.11 times faster than the state-of-the-art solution. Our results also show that the shortest guaranteed-cost path can reduce the cost of the recovery path significantly. Moreover, we design a prototype to show how to deploy our algorithm in an OpenFlow network.

With an increasing demand for flexible management in software-defined networks (SDNs), it becomes critical to minimize the network policy update time. Although major SDN controllers are now optimized for rapid network update at the control plane, there is still room for data plane optimization in terms of update time, when using TCAM-based physical SDN commodity-off-the-shelf switches. A slow update directly affects network performance creating bottlenecks. To minimize flow entry update time, a dependency graph, a kind of DAG (directed acyclic graph), can be used for the access management of flow entries at the switch. Thanks to the DAG, unnecessary entry movements, which are the main factor slowing down flow entry updates, can be avoided. However, existing algorithms show limitations when updates become very frequent. We propose a new flow entry update algorithm, called FastRule, that exploits a greedy strategy with an efficient data structure to accelerate flow entry update with a DAG approach. Moreover, we also adjust our algorithm for other flow table layouts to make it scalable. We elaborate on the correctness of FastRule and test our algorithm using a hardware switch. Compared with existing algorithms, the evaluation shows that our algorithm is about 100x faster than state-of-the-art solutions with a flow table of 1k line size.

Heterogeneous wireless networks operate under un-certainties as traffic load variation and failures. Redundant packet scheduling algorithms seek to timely cope with these uncertainties to meet the latency and throughput requirements for delay-sensitive applications, such as online gaming, voice and video streaming. The Multipath Transmission Control Protocol (MPTCP) can rely on a redundant scheduler to fulfill this demand, replicating data packets on multiple paths to mitigate negative effects of heterogeneity, such as fluctuations in delay, loss rate, and bandwidth. However, shared bottlenecks among paths can compromise the benefits of the redundant scheduler and restrain the MPTCP performance. Thus, this paper introduces a new scheduler, called RED (REdundant Diversity scheduling), to prioritize packet replication by uncorrelated paths. RED selects paths and replicates packets based on the Spearman's correlation coefficient. Results show that RED achieves low delay and enhances the usage of low correlated paths, improving transmission performance.

Mobile devices are equipped with limited processing power and battery charge. A mobile computation offloading framework is a software that provides better user experience in terms of computation time and energy consumption, also taking profit from edge computing facilities. This article presents User-Level Online Offloading Framework (ULOOF), a lightweight and efficient framework for mobile computation offloading. ULOOF is equipped with a decision engine that minimizes remote execution overhead, while not requiring any modification in the device’s operating system. By means of real experiments with Android systems and simulations using large-scale data from a major cellular network provider, we show that ULOOF can offload up to 73 percent of computations, and improve the execution time by 50 percent while at the same time significantly reducing the energy consumption of mobile devices.

In networking and computing, resource allocation is typically addressed using classical resource allocation protocols as the proportional rule, the max-min fair allocation, or solutions inspired by cooperative game theory. In this paper, we argue that, under awareness about the available resource and other users demands, a cooperative setting has to be considered in order to revisit and adapt the concept of fairness. Such a complete information sharing setting is expected to happen in 5G environments, where resource sharing among tenants (slices) need to be made acceptable by users and applications, which therefore need to be better informed about the system status via ad-hoc (northbound) interfaces than in legacy environments. We identify in the individual satisfaction rates the key aspect of the challenge of defining a new notion of fairness in systems with complete information sharing, consequently, a more appropriate resource allocation algorithm. We generalize the concept of user satisfaction considering the set of admissible solutions for bankruptcy games and we adapt to it the fairness indices. Accordingly, we propose a new allocation rule we call mood value: for each user, it equalizes our novel game-theoretic definition of user satisfaction with respect to a distribution of the resource. We test the mood value and a new fairness index through extensive simulations about the cellular frequency scheduling use-case, showing how they better support the fairness analysis. We complete the paper with further analysis on the behavior of the mood value in the presence of multiple competing providers and with cheating users.

Multipath communications at the Internet scale have been a myth for a long time, with no actual protocol being deployed so that multiple paths could be taken by a same connection on the way towards an Internet destination. Recently, the Multipath Transmission Control Protocol (MPTCP) extension has been standardized and is undergoing rapid adoption in many different use-cases, from mobile to fixed access networks, from data-centers to core networks. Among its major benefits - i.e., reliability thanks to backup path rerouting, throughput increase thanks to link aggregation, and confidentiality being more difficult to intercept a full connection - the latter has attracted lower attention. How effective would be to use MPTCP to exploit multiple Internet-scale paths and decrease the probability of Man-in-the-Middle (MITM) attacks is a question which we try to answer. By analyzing the Autonomous System (AS) level graph, we identify which countries and regions show a higher level of robustness against MITM AS-level attacks, for example due to core cable tapping or route hijacking practices.

Mobile computation offloading has recently attracted much interest and first offloading solutions have been developed. However, the relevant technical challenge of how to automatically determine offloadable sections of Android applications has not been adequately investigated so far. This paper proposes an innovative task selection algorithm that can parse an Android application autonomously and classify all the methods based on their offloadability by adopting a fine grained and multi-steps analyzer. The reported experimental results show the effectiveness of our solution when applied to the top 25 most downloaded Android apps on the Google Play store, by showing its accuracy in identifying off loadable methods and demonstrating the potential benefits of automated mobile computation offloading.

In this paper we discuss the use of Bayesian network graphs for network modeling and resiliency estimation problems related to Private Mobile Radio (PMR) networks. We specifically focus on network availability computation and fault diagnosis, when measurement data of only a part of the system is available to infer the state of the rest of the network. The resiliency of such uncertain networks strongly depends on the adopted redundancy mechanism. Based on bayesian graph and probability propagation techniques, we model and evaluate PMR system availability using different redundancy schemes. Our results show that the bayesian network model we propose provides accurate and straightforward estimations on system reliability and fault diagnosis.

The fundamental tasks of the control plane in software defined networking (SDN) are to customize forwarding policies for the data plane and to provide global network view for applications. The logically centralized control plane design brings benefits in terms of network programmability and can largely ease network management. However, it also increases efficiency concerns. One practical control plane challenge is path computation, because it can require a significant amount of computation load if the network scale is large and the path requests from applications are frequent. In this paper, our goal is to build a high-performance control plane for path computation using multiple controllers. Previous works attempt to improve control plane efficiency by balancing only the load for data plane behavior between multiple controllers. Going beyond conventional wisdom, we designed ParaCon, a solution we propose to speed up the control plane by distributing the load of path computation. We also address the consistency and synchronization overhead challenges related to ParaCon design. To the best of our knowledge, ParaCon is the first attempt that utilizes node parallelism in SDN path computation. We evaluated ParaCon using both Mininet and real-world clusters. Our results show that the path computing time of ParaCon can achieve a speedup of 10× over Floyd (used in POX) and Dijkstra (used in ONOS) baseline implementations for networks with hundreds of nodes.

Major interest is currently given to the integration of clusters of virtualization servers, also referred to as `cloudlets' or `edge clouds', into the access network to allow higher performance and reliability in the access to mobile edge computing services. We tackle the edge cloud network design problem for mobile access networks. The model is such that the virtual machines (VMs) are associated with mobile users and are allocated to cloudlets. Designing an edge cloud network implies first determining where to install cloudlet facilities among the available sites, then assigning sets of access points, such as base stations to cloudlets, while supporting VM orchestration and considering partial user mobility information, as well as the satisfaction of service-level agreements. We present link-path formulations supported by heuristics to compute solutions in reasonable time. We qualify the advantage in considering mobility for both users and VMs as up to 20% less users not satisfied in their SLA with a little increase of opened facilities. We compare two VM mobility modes, bulk and live migration, as a function of mobile cloud service requirements, determining that a high preference should be given to live migration, while bulk migrations seem to be a feasible alternative on delay-stringent tiny-disk services, such as augmented reality support, and only with further relaxation on network constraints.

The rapid growth of mobile traffic and the emergence of advanced mobile services and infrastructures are shifting significant attention toward the cellular network back-hauling infrastructure. At this network segment, there is a growing interest in understanding spatio-temporal mobile traffic distributions at different network levels, in order to better define flexible networking solutions for forthcoming smart 5G infrastructures including, for instance, mobile edge computing features. In this work we study these aspects and characterize the load on cellular access networks using real-world anonymized subscriber data, from the Lyon metropolitan area in France, providing statistical distribution to the research community. We find that the traffic distribution at Node-B level is best fit by a Weibull distribution, and that at the radio network aggregation it is best fit by a hybrid Weibull-Pareto distribution.

Nowadays, the explosion of cloud-based applications is leading to a much higher demand on both computing and network infrastructure resources than only a few years ago. Enhancing the user experience, by reducing the latency and increasing network stability, becomes an important challenge for cloud operators. In this paper, we propose a unified protocol architecture, based on the Locator/Identifier Separation Protocol (LISP) and the Transparent Interconnection of a Lots of Links (TRILL) protocol, to enhance the access performance and to minimize the retrieval latency for services hosted in a distributed data center (DC) fabric. LISP is used as a cloud access overlay protocol, while TRILL is used as a geo-distributed DC virtual network overlay protocol. We specify and design a cross-layer protocol agent able to map virtual network embedding information from TRILL (layer 2) to LISP (layer 3) in order to allow cloud providers to let the client access the cloud by the best DC entry point, assuming that the inter-DC site latency dominates over the DC access latency. We tested our architecture in a real testbed. We compared the proposed solution to the legacy situation, highlighting the achievable gains in cloud access latency.

In this paper, we propose a new approach for modeling and evaluating reliability of complex public safety network. Based on Bayesian network structure, we develop a general construction methodology for network representation and design. Using probability propagation techniques, the reliability of the system can be obtained in a straightforward manner. To evaluate our model, the proposed method is illustrated through three PSN scenarios that exemplify real emergency situations.

Cloud applications heavily rely on the network communication infrastructure, whose stability and latency directly affect the quality of experience. As mobile devices need to rapidly retrieve data from the cloud, it becomes an extremely important goal to deliver the lowest possible access latency at the best reliability. In this paper, we specify a cloud access overlay protocol architecture to improve the cloud access performance in distributed data-center (DC) cloud fabrics. We explore how linking virtual machine (VM) mobility and routing to user mobility can compensate performance decrease due to increased user-cloud network distance, by building an online cloud scheduling solution to optimally switch VM routing locators and to relocate VMs across DC sites, as a function of user-DC overlay network states. We evaluate our solution: 1) on a real distributed DC testbed spanning all of France, showing that we can grant a very high transfer time gain and 2) by emulating the situation of Internet service providers (ISPs) and over-the-top (OTT) cloud providers, exploiting thousands of real France-wide user displacement traces, finding a median throughput gain from 30% for OTT scenarii to 40% for ISP scenarii, the large majority of this gain being granted by adaptive VM mobility.

The use by a growing number of users of Cloud-based services requires an adaptation of the network technologies used to access them. We propose to combine two novel protocols at the state of the art at Cloud access middle-boxes to better profit from spare unused network path diversity. The first protocol, Multipath TCP, allows creating multiple TCP/IP sub flows, as much as needed. The second, the Locator/Identifier Separation Protocol (LISP), can be used to route the subflows on different wide-area network paths, possibly disjoint, and also allows native support for seamless virtual machine migrations. In this paper we specify how we can combine these two protocols to increase the bandwidth available to access applications run in multi-homed data-centers. We describe how these protocols can be integrated into a Cloud access middle-box. By means of a combined MPTCP-LISP access proxy, the acceleration is transparent to the user terminal that does not necessitate any upgrade. We provide the detailed system-level architecture based on open source code, and we document results from preliminary experimentations on one of two targeted use-cases. The evaluations conducted show that the overhead generated by our solution remains moderate despite the various system-level steps required to translate incoming TCP packets into MPTCP-LISP packets then routed over different IP paths.

Several works in the literature show that accurate estimations of the actual One-Way Delays (OWD) could improve the performance of various network protocol, such as Transport Control Protocol (TCP) throughput. With the emergence of multipath transport protocols like Multipath TCP or the Stream Control Transport Protocol (SCTP), the potential impact can be even higher. Indeed, as multipath transport protocols send data concurrently on heterogeneous paths, the knowledge of corresponding OWDs can greatly help mitigating packet arrival disorder. Theoretically, clock synchronization protocols between endpoints could ensure OWD knowledge, but their efficiency at the Internet scale is debatable. In practice, TCP uses the Round Trip Time (RTT) to take into account congestion or to compute retransmission timeouts, and the OWD is assumed to be half the RTT. However, numerous studies show that a majority of Internet connections experience latency asymmetry. In this paper, we propose a technique based on differential pacing over multiple paths to obtain an estimation of the difference in OWDs between the different paths, motivated by its strong utility for multipath transport protocols such as MPTCP and SCTP. Simulations show which are the interesting scenarios of application.

In spite of their rapid growth, cloud applications still heavily rely on the network communication infrastructure, whose stability and latency directly affect the quality of experience. In fact, as mobile devices need to rapidly get real-time information and files from the cloud, it becomes an extremely important factor for cloud providers to deliver a better user experience. In this paper, we specify a cloud access overlay protocol architecture, based on traffic engineering extensions of the Locator/Identifier Separation Protocol (LISP), to improve the access performance for Cloud services delivered by a distributed data center fabric. The distributed fabric offers the possibility to access the services through multiple routing locators and to migrate server virtual machines (VMs) to different locations improving access performance. We address the problem of jointly switching VM routing locators and migrating VMs across data-center sites. We propose an adaptive control framework that allows satisfying agreed-upon levels of quality of service. We evaluate the architecture on a real distributed data-center network, involving four distant LISP-enabled data-center sites in France, as compared to legacy situations with no Cloud access optimization. By emulating realistic situations we show that, by only switching the data-center routing locator, we can guarantee a better user experience with a transfer time decreased by 80%. Moreover, we show that, to react to situations when the Cloud access link between sites is disrupted or suffers excessively from packet loss, the adaptive VM migration policy can further decrease the transfer time by 40%.

Major interest is currently given to the integration of clusters of virtualization servers, also referred to as `cloudlets', into the access network to allow higher performance and reliability in the access to mobile cloud services. We tackle the cloudlet network design problem for mobile access networks. The model is such that virtual machines are associated with mobile users and are allocated to cloudlets. Designing a cloudlet network implies first determining where to install cloudlet facilities among the available sites, then assigning sets of access points such as base-stations to cloudlets, while supporting virtual machine migrations and taking into account partial user mobility information, as well as the satisfaction of service-level agreements. We present link-path formulations supported by heuristics to compute solutions in reasonable time. We qualify the advantage in considering mobility for both users and virtual machines as up to 40% less cloudlet facilities to install and 40% less virtual machine migrations to execute. We compare two migration modes, bulk and live migration, as a function of mobile cloud service requirements, determining that a high preference should be given to bulk migrations for delay-stringent services such as augmented reality support, while for applications with less stringent delay requirements, live migration appears as largely preferable.

Femtocells are emerging as a key technology to improve coverage and network capacity in indoor environments. When femtocells use different frequency bands than macrocells (i.e., split-spectrum approach), femto-to-femto interference remains the major issue. In particular, congestion cases in which femtocell demands exceed the available resources raise several challenging questions: how much a femtocell can demand? how much it can obtain? and how this shall depends on the interference with its neighbors? Strategic interference management between femtocells via power control and resource allocation mechanisms is needed to avoid performance degradation during congestion cases. In this paper, we model the resource and power allocation problem as an operations research game, where imputations are deduced from cooperative game theory, namely the Shapley value and the Nucleolus, using utility components results of partial optimizations. Based on these evaluations, users' demands are first rescaled to strategically justified values. Then, a power-level and throughput optimization using the rescaled demands is conducted. The performance of the developed solutions is analyzed and extensive simulation results are presented to illustrate their potential advantages. In particular, we show that the Shapley value solution with power control offers the overall best performance in terms of throughput, fairness, spectrum spatial reuse, and transmit power, with a slightly higher time complexity compared to alternative solutions.

The increasing adoption of server virtualization has recently favored three key technology advances in data-center networking: the emergence at the hypervisor software level of virtual bridging functions between virtual machines and the physical network; the possibility to dynamically migrate virtual machines across virtualization servers in the data-center network (DCN); a more efficient exploitation of the large path diversity by means of multipath forwarding protocols. In this paper, we investigate the impact of these novel features in DCN optimization by providing a comprehensive mathematical formulation and a repeated matching heuristic for its resolution. We show, in particular, how virtual bridging and multipath forwarding impact common DCN optimization goals, traffic engineering (TE) and energy efficiency (EE), and assess their utility in the various cases of four different DCN topologies. We show that virtual bridging brings a high performance gain when TE is the primary goal and should be deactivated when EE becomes important. Moreover, we show that multipath forwarding can bring relevant gains only when EE is the primary goal and virtual bridging is not enabled.

A hot topic in data center design is to envision geo-distributed architectures spanning a few sites across wide area networks, allowing more proximity to the end users and higher survivability, defined as the capacity of a system to operate after failures. As a shortcoming, this approach is subject to an increase of latency between servers, caused by their geographic distances. In this paper, we address the trade-off between latency and survivability in geo-distributed data centers, through the formulation of an optimization problem. Simulations considering realistic scenarios show that the latency increase is significant only in the case of very strong survivability requirements, whereas it is negligible for moderate survivability requirements. For instance, the worst-case latency is less than 4 ms when guaranteeing that 80% of the servers are available after a failure, in a network where the latency could be up to 33 ms.

A present issue in the evolution of mobile cellular networks is determining whether, how and where to deploy adaptive content and cloud distribution solutions at the base station and backhauling network level. Intuitively, an adaptive placement of content and computing resources in the most crowded regions can grant important traffic offloading, improve network efficiency and user quality of experience. In this paper we document the content consumption in the Orange cellular network for the Paris metropolitan area, from spatial and application-level extensive analysis of real data from a few million users, reporting the experimental distributions. In this scope, we propose a hotspot cell estimator computed over user's mobility metrics and based on linear regression. Evaluating our estimator on real data, it appears as an excellent hotspot detection solution of cellular and backhauling network management. We show that its error strictly decreases with the cell load, and it is negligible for reasonable hotspot cell load upper thresholds. We also show that our hotspot estimator is quite scalable against mobility data volume and against time variations.

Nowadays, the rapid growth of Cloud computing services is stressing the network communication infrastructure in terms of resiliency and programmability. This evolution reveals missing blocks of the current Internet Protocol architecture, in particular in terms of virtual machine mobility management for addressing and locator-identifier mapping. In this paper, we propose some changes to the Locator/Identifier Separation Protocol (LISP) to cope with this gap. We define novel control-plane functions and evaluate them exhaustively in the worldwide public LISP testbed, involving five LISP sites distant from a few hundred kilometers to many thousands kilometers. Our results show that we can guarantee service downtime upon live virtual machine migration lower than a second across American, Asian and European LISP sites, and down to 300 ms within Europe, outperforming standard LISP and legacy triangular routing approaches in terms of service downtime, as a function of datacenter-datacenter and client-datacenter distances.

Today's mobile terminals have several access network interfaces. New protocols have been proposed during the last few years to enable the concurrent use of multiple access paths for data transmission. In practice, the use of different access technologies is subject to different interconnection costs, and mobile users have preferences on interfaces jointly depending on performance and cost factors. There is therefore an interest in defining “light” multipath communication policies that are less expensive than greedy unconstrained ones such as with basic multipath TCP (MP-TCP) and that are strategically acceptable assuming a selfish endpoint behavior. With this goal, we analyze the performance-cost trade-off of multi-homed end-to-end communications from a strategic standpoint. We model the communication between multi-homed terminals as a specific non-cooperative game to achieve performance-cost decision frontiers. The resulting potential game always allows selecting multiple equilibria, leading to a strategic load-balancing distribution over the available interfaces, possibly constraining their use with respect to basic MP-TCP. By simulation of a realistic three-interface scenario, we show how the achievable performance is bound by the interconnection cost; we show that we can halve the interconnection cost with respect to basic (greedy) MP-TCP while offering double throughputs with respect to single-path TCP. Moreover, we evaluate the compromise between keeping or relaxing strategic constraints in a coordinated MP-TCP context.

In this paper, we present a generalized multipath load sharing framework for Internet. Under the hypothesis that edge networks are fully independent in establishing Internet routing policies, we use game theory and a vectorized utility function to solve the Internet multipath routing problem. We first briefly review the existing techniques that apply game theory to improve the Internet routing resiliency. We propose a vectorized routing cost model to consider multiple routing metrics in the network setting, along with a universal refinement method to quantify the profile performance and predict the behavior of the vectorized routing game. Based on the universal refinement method as well as a linear traffic distribution algorithm, we define a generalized multipath load sharing framework to improve the routing resiliency in traffic exchange between two distant edge networks. Running simulations with real measured Internet metrics, we find that the proposed generalized edge-to-edge multipath load sharing framework is able to offer far more resilient solutions than legacy and alternative protocols such as standard BGP and BGP with LISP-based traffic engineering.

Femtocell provisioning is emerging as a key technology to improve coverage and network capacity in indoor environments. When femtocells use different frequency bands than macrocells (i.e., split-spectrum approach), femto-to-femto interference remains the major issue. In particular, congestion cases in which femtocell demands exceed the available resources pose an important challenge. In this paper, we propose a joint resource and power allocation strategy for the management of interference in cooperative femtocell networks. We model the resource and power allocation problem as an operations research game, where imputations are deduced from cooperative game theory, namely the Shapley value and the Nucleolus, using utility components results of partial optimizations. The performance of the developed solutions is analyzed and extensive simulation results are presented to illustrate their potential advantages. In particular, we show that the Shapley value solution with power control offers the overall best performance in terms of throughput, fairness, and transmit power, compared to alternative solutions.

Cloud networking imposes new requirements in terms of connection resiliency and throughput among virtual machines, hypervisors and users. A promising direction is to resort to multipath communications, yet existing protocols still struggle to take advantage of the path diversity offered by IP networks. Multipath TCP (MPTCP) can create several TCP subflows on different interfaces and concurrently forward data on these subflows. Current MPTCP implementations create a full mesh of subflows between IP endhosts, which may be suboptimal according to the topology. We propose to rely on topology information brought by an external protocol in order to improve the MPTCP subflow management; we resort to the Locator/Identifier Separation Protocol (LISP) to retrieve IP path diversity information, to then accordingly create MPTCP subflows. We report noticeable benefits obtained using a large-scale Cloud access test bed, and we describe further work we are conducting in this sense.

Nowadays, the huge worldwide mobile-phone penetration is increasingly turning the mobile network into a gigantic ubiquitous sensing platform, enabling large-scale analysis and applications. In recent years, mobile data-based research reaches important conclusions about various aspects of human mobility patterns and trajectories. But how accurately do these conclusions reflect the reality? In order to evaluate the difference between the reality and the approximation methods, we study in this paper the error between real human trajectory and the one obtained through mobile phone data using different interpolation methods (linear, cubic, nearest and spline interpolations) while taking into account some mobility parameters. From extensive evaluations based on real cellular network activity data of the Boston metropolitan area, we show that the linear interpolation offers the best estimation for sedentary people and the cubic one for commuters. Moreover, the nearest interpolation appears as the best one for “ordinary people” doing regular stops and standard displacements. Another important experimental finding described in this paper is that trajectory estimation methods show different error regimes whether used within or outside the “territory” of the user defined by the radius of gyration.

Wireless mesh networks (WMNs) are emerging as a key solution to provide broadband and mobile wireless connectivity in a flexible and cost-effective way. In suburban areas, a common deployment model relies on orthogonal frequency division multiple access (OFDMA) communications between mesh routers (MRs), with one MR installed at each user premises. In this paper, we investigate a possible user cooperation path to implement strategic resource allocation in OFDMA WMNs, under the assumption that users want to control their interconnections. In this case, a novel strategic situation appears: How much an MR can demand, how much it can obtain, and how this shall depend on the interference with its neighbors. Strategic interference management and resource allocation mechanisms are needed to avoid performance degradation during congestion cases between MRs. In this paper, we model the problem as a bankruptcy game taking into account the interference between MRs. We identify possible solutions from cooperative game theory, namely the Shapley value and the nucleolus, and show through extensive simulations of realistic scenarios that they outperform two state-of-the-art OFDMA allocation schemes, namely, centralized-dynamic frequency planning, and frequency-ALOHA. In particular, the nucleolus solution offers best performance overall in terms of throughput and fairness, at a lower time complexity.

Femtocells have recently appeared as a viable solution to enable broadband connectivity in mobile cellular networks. Instead of redimensioning macrocells at the base station level, the modular installation of short-range access points can grant multiple benefits, provided that interference is efficiently managed. In the case where femtocells use different frequency bands than macrocells (i.e., split-spectrum approach), interference between femtocells is the major issue. In particular, congestion cases in which femtocell demands exceed the available bandwidth pose an important challenge. If, as expected, the femtocell service is going to be separately billed by legacy wire-line Internet Service Providers, strategic interference management and resource allocation mechanisms are needed to avoid performance degradation during congestion cases. In this paper, we model the resource allocation in cooperative femtocell networks as a bankruptcy game. We identify possible solutions from cooperative game theory, namely the Shapley value and the Nucleolus, and show through extensive simulations of realistic scenarios that they outperform two state-of-the-art schemes, namely Centralized-Dynamic Frequency Planning, C-DFP, and Frequency-ALOHA, F-ALOHA. In particular, the Nucleolus solution offers best performance overall in terms of throughput and fairness, at a lower time complexity.

Wireless Mesh Networks (WMNs) are emerging as a key solution to provide broadband and mobile wireless connectivity in a flexible and cost effective way. In suburban areas, a common deployment model relies on OFDMA communications between mesh routers (MRs), with one MR installed at each user premises. In this paper, we investigate a possible user cooperation path to implement strategic resource allocation in OFDMA WMNs, under the assumption that users want to control their interconnection. In this case, a novel strategic situation appears: how much a MR can demand, how much it can obtain and how this shall depend on the interference with its neighbors. Strategic interference management and resource allocation mechanisms are needed to avoid performance degradation during congestion cases between MRs. In this paper, we model the problem as a bankruptcy game taking into account the interference between MRs. We identify possible solutions from cooperative game theory, namely the Shapley value and the Nucleolus, and show that they outperform two state-of-the-art schemes, namely Centralized-Dynamic Frequency Planning, C-DFP, and Frequency-ALOHA, F-ALOHA. In particular, the Nucleolus solution offers best performance overall in terms of throughput and fairness.

Today's mobile terminals have several access network interfaces. In practice, the use of different access technologies is subject to different interconnection costs, and mobile users have preferences on interfaces jointly depending on performance and cost factors. There is therefore an interest in defining “light” yet rational multipath communication policies less expensive than greedy ones such as with basic Multipath TCP (MP-TCP). We analyze the performance-cost trade-off of multi-homed end-to-end communications from a strategic standpoint. We model the communication between multi-homed terminals as a multi-criteria non-cooperative game so as to achieve performance-cost decision frontiers. The resulting potential game always allows to select multiple equilibria, which correspond to a strategic load-balancing distribution over the available interfaces, possibly constraining their use with respect to basic MP-TCP. We specify how the resulting model may be in practice implemented by users willing to jointly control the interconnection cost and the performance, based on user Quality of Experience (QoE) assessments. By simulation of a realistic 3-interface scenario, we show how the achievable performance is bound by the interconnection cost; we show that we can halve the interconnection cost with respect to basic (greedy) MP-TCP under a reasonable trade-off, while offering double throughputs with respect to single-path TCP.

The significant growth in the global Internet traffic and routing table size requires solutions to address Internet scalability and resiliency. A number of proposals have considered moving away from the flat legacy Internet routing to a two-level hierarchical routing, separating edge networks from transit carrier networks. In this paper, we study the extended inter-domain traffic engineering capabilities arising in a transit-edge separated Internet routing, focusing on those multi-homed edge networks (e.g., small ISPs, content providers, large corporations) that aim at increasing their Internet resiliency experience. We model using game theory the interaction between distant independent edge networks exchanging large traffic volumes, with the goal of seeking efficient edge-to-edge load-balancing routing solutions. The proposed traffic engineering framework relies on a non-cooperative potential game, built upon path prepending- and path diversity- dependent costs, that indicates efficient equilibrium solution for the edge-to-edge load-balancing coordination problem. Simulations on real instances show that, in comparison with the alternative multipath BGP and normal LISP, we can achieve significantly higher resiliency and stability. In particular, our simulation for an illustrating case shows four-times more stable multipath routing solutions with a five-times larger path diversity.

We present a novel resilient routing policy for controlling the routing across peering links between Internet carriers. Our policy is aimed at offering more dependability and better performance to the routing decision with respect to the current practice (e.g., hot-potato routing). Our work relies on a non-cooperative game framework, called Peering Equilibrium MultiPath (PEMP), that has been recently proposed. PEMP allows two carrier providers to coordinate a multipath route selection for critical flows across peering links, while preserving their respective interests and independence. In this paper, we propose a resilient PEMP execution policy accounting for the occurrence of potential impairments (traffic matrix variations, intra-AS and peering link failures) that may occur in both peering networks. We mathematically define how to produce robust equilibrium sets and describe how to appropriately react to unexpected network impairments that might take place. The results from extensive simulations show that, under a realistic failure scenario, our policy adaptively prevents from peering link congestions and excessive route deviations after failures.

Abstract—It is generally admitted that interdomain peering links represent nowadays the main bottleneck of the Internet, particularly because of lack of coordination between providers, which use independent and “selfish ” routing policies. We are interested in identifying possible “light ” coordination strategies that would allow carriers to better control their peering links while preserving their independence and respective interests. We propose a robust multipath routing coordination framework for peering carriers, which relies on the multiple-exit discriminator (MED) attribute of Border Gateway Protocol (BGP) as signaling medium. Our scheme relies on a game theory modeling, with a non-cooperative potential game considering both routing and congestions costs. Peering equilibrium multipath (PEMP) coordination policies can be implemented by selecting Pareto-superior Nash equilibria at each carrier. We compare different PEMP policies to BGP Multipath schemes by emulating a realistic peering scenario. Our results show that the routing cost can be decreased by roughly 10 % with PEMP. We also show that the stability of routes can be significantly improved and that congestion can be practically avoided on the peering links. Finally, we discuss practical implementation aspects and extend the model to multiple players highlighting the possible incentives for the resulting extended peering framework. Index Terms—Border Gateway Protocol (BGP), game theory, interdomain routing, multiple-exit discriminator (MED), multipath, peering. I.

We present a novel resilient routing policy for controlling the routing across peering links between Internet carriers. Our policy is aimed at offering more reliability, dependability and better performance to the routing decision with respect to the current practice (e.g., hot-potato routing). Our work relies on a non-cooperative game framework, called Peering Equilibrium MultiPath (PEMP), that has been recently proposed. PEMP allows two carrier providers to coordinate a multipath route selection for critical flows across peering links, while preserving their respective interests and independence. In this paper, we propose a resilient PEMP execution policy accounting for the occurrence of potential impairments (traffic matrix variations, intra-AS and peering link failures) that may occurr in both peering networks. We mathematically define how to produce robust equilibrium sets and describe how to appropriately react to unexpected network impairments that might take place. The results from extensive simulations show that, under a realistic failure scenario, our policy adaptively prevents from peering link congestions and excessive route deviations after failures.

It is generally admitted that Inter-domain peering links represent nowadays the main bottleneck of the Internet, particularly because of lack of coordination between providers, which use independent and "selfish" routing policies. We are interested in identifying possible "light" coordination strategies, that would allow carriers to better control their peering links, while preserving their independence and respective interests. We propose a robust multi-path routing coordination framework for peering carriers, which relies on the MED attribute of BGP as signalling medium. Our scheme relies on a game theoretic modelling, with a non-cooperative potential game considering both routing and congestions costs. Peering Equilibrium MultiPath (PEMP) coordination policies can be implemented by selecting Pareto-superior Nash equilibria at each carrier. We compare different PEMP policies to BGP Multipath schemes by emulating a realistic peering scenario. Our results show that the routing cost can be decreased by roughly 10% with PEMP. We also show that the stability of routes can be significantly improved and that congestion can be practically avoided on the peering links.

In this paper, we study the problem of inter-domain AS tree selection for multipoint tunnel set-up within an alliance of ASs. We first describe the framework of our work, based on the introduction of a service plane for automatic multi-domain service provisioning. We introduce an abstract representation of domain relationship by means of directional metrics which are applied to a triplet (ingress point, transit AS, egress point) where the ingress and egress points can be ASs or routers. Then, we focus on the multipoint AS Selection problem that arises in such an architecture. The corresponding constrained Steiner problem is known to be a hard problem, and the introduction of directional metrics increases its complexity. We propose an original approach that allows one to reach almost optimal solutions with tractable computation times. Besides its performance, one contribution of this paper is that some steps of the proposed heuristic can be precomputed, independently of the tunnel demands. By extensive tests on random topologies derived from the Internet, we show that our heuristic is often equal or a few percent close to the optimal, and that, in the case of precomputation, its time consumption can be much lower than other well-known algorithms.

This paper presents the design and dimensioning optimization of a novel optical network structure, called the Petaweb, having a total capacity of several Pb/s (1015bit/s). Its topology is a superimposition of stars that drastically eases signaling and switching operations. Firstly, we deal with the network model, focusing on the insertion of the time-sharing of the optical channel in network components and in lightpath provisioning. The design problem is jointly a network dimensioning and an assignment problem; we propose for the dimensioning an integer linear programming formulation and a linear resolution algorithm for the assignment. We also propose the use of a quasi-regular topology extracted from the optimized regular topology to reduce costs and improve the network utilization.