Radio access network (RAN) slicing in unmanned aerial vehicle (UAV)-assisted tactical networks facilitates dynamic sharing of radio resources among slices with varying quality of service (QoS) requirements. However, the unpredictable and resource-constrained nature of tactical RAN environments can jeopardize slice isolation, significantly impacting the QoS provided to users. This paper introduces a UAV positioning and bandwidth allocation (UAV-PBA) mechanism that optimally positions UAVs to enhance coverage and allocates bandwidth among users while considering slice isolation and user mobility. UAV-PBA ensures robust slice isolation by satisfying both resource-based and performance-based constraints. The mechanism utilizes deep reinforcement learning (DRL) algorithms for effective UAV positioning and bandwidth allocation. We design two implementations of the UAV-PBA mechanism, each utilizing a different DRL algorithm, and evaluate their performance against baseline methods using several metrics.

Thanks to its ability to provide secure and efficient communication capabilities, network slicing has been adopted as a key technology, not just in commercial networks but also in tactical networks. Based on multiple virtual networks, each dedicated to a specific service, a sliced architecture meets the diverse requirements of highly heterogeneous tactical services. However, tactical networks operate in challenging environments with inherent power constraints, where the need for energy-efficient network slicing management solutions is paramount. In this direction, we tackle a joint slice activation/deactivation and user association problem with the aim of studying trade-offs between energy efficiency and user quality of service. To solve the problem, we introduce our original approach: a split reinforcement learning-based energy-efficient slicing deployment algorithm, namely SNOW. SNOW divides the deep neural network into multiple sections, where the front-end part of the model is trained over multiple user devices and then the back-end part of the model is trained by the central nodes (i.e. base stations in this case), without sensitive data sharing. Extensive simulation results reveal that the proposed scheme is superior to the considered benchmarks in improving energy efficiency while maintaining network performance.

Network slicing is one of the major catalysts proposed to turn future telecommunication networks into versatile service platforms. Along with its benefits, network slicing is introducing new challenges in the development of sustainable network operations, as it entails a higher energy consumption compared to non-sliced networks.Using a sliced architecture, which includes guaranteeing the communication and computation requirements for each slice, is essential for operators to provide a satisfying user quality of service (QoS) in a multi-service network. At the same time, building sustainable mobile networks, with the least amount of resources used, is crucial today, for both economic and environmental reasons. As a result, mobile operators need to find a middle ground between these two objectives – a tough nut considering they are both antithetical and important. In this light, we investigate a joint slice activation/deactivation and user association problem, with the aim of minimizing energy consumption and maximizing the QoS. The proposed multI-agent fully CooperativE deCentRalizEd frAMework (ICE-CREAM) addresses the formulated joint problem, with agents acting at two different granularity levels. Not only all the agents can access the shared information with their direct neighbors, but also they are trained with one global reward, which is an ideal approach in multi-agent cooperative settings. We evaluate ICE-CREAM using a real-world dataset that captures the spatio-temporal consumption of three different mobile services in France. Experimental results demonstrate that the proposed solution provides more than 30% energy efficiency improvement compared to a configuration where all the slice instances are always active while maintaining the same level of QoS. From a broader perspective, our work explicitly shows the impact of prioritizing the energy over QoS, and vice versa.

The futuristic sixth-generation (6-G) networks will empower ultrareliable and low latency communications (URLLC), enabling a wide array of mission-critical applications such as mobile edge computing (MEC) systems, which are largely unsupported by fixed communication infrastructure. To remedy this issue, unmanned aerial vehicle (UAV) has recently come to the limelight to facilitate MEC for internet of things (IoT) devices as they provide desirable line-of-sight (LoS) communications compared to fixed terrestrial networks, thanks to their added flexibility and 3-D positioning. In this article, we consider UAV-enabled relaying for MEC systems for uplink transmissions in 6-G networks, and we aim to optimize mission completion time subject to the constraints of resource allocation, including UAV transmit power, UAV CPU frequency, decoding error rate, blocklength, communication bandwidth, and task partitioning as well as 3-D UAV positioning. Moreover, to solve the nonconvex optimization problem, we propose three different algorithms, including successive convex approximations, altered genetic algorithm (AGA), and smart exhaustive search. Thereafter, based on time-complexity, execution time, and convergence analysis, we select AGA to solve the given optimization problem. Simulation results demonstrate that the proposed algorithm can successfully minimize the mission completion time, perform power allocation at the UAV side to mitigate information leakage and eavesdropping as well as map a 3-D UAV positioning, yielding better results compared to the fixed benchmark submethods. Lastly, subject to 3-D UAV positioning, AGA can also effectively reduce the decoding error rate for supporting URLLC services.

Jamming attacks are becoming increasingly common and pose a significant threat to the security and reliability of wireless sensor networks (WSNs). These attacks can be difficult to detect, as they often operate in a stealthy manner, disrupting communication between sensors. Artificial intelligence (AI) techniques have the potential to be highly effective in detecting jamming attacks. However, the adoption of AI-based techniques for detecting jamming attacks has been limited due to the scarcity of up-to-date and accurate data of these attacks. Privacy-aware collaboration among agents is expected to be essential in building robust AI-based models for detecting jamming attacks in WSNs. In this context, we propose a novel framework that uses collaborative federated learning (FL) to enable privacy-aware distributed learning among multiple agents without sharing their sensitive data. This can be particularly important in jamming attack detection, where data may contain sensitive information that should not be shared. In addition, we design a novel secure aggregation scheme to protect the FL aggregation service from reverse-engineering attacks. The effectiveness of our proposed framework was tested using the public Wireless Sensor Networks Dataset (WSN-DS) that includes four types of well-known jamming attacks (i.e., constant jamming, reactive jamming, random jamming, and deceptive jamming). The results of a thorough experiment using the WSN-DS data set demonstrate the effectiveness and high accuracy/F1-score (99%) in detecting jamming attacks while also maintaining participant privacy.

Software-oriented architecture (SOA) is a promising paradigm for efficiently leveraging the functionality of individual IoT devices to build IoT applications. However, deploying SOA for IoT data-gathering applications requires spatial context-awareness and the ability to aggregate similar available services, which presents a challenge. To address this challenge, this paper proposes a formulation for spatial context-aware service composition with a novel quantitative model for spatial context. We demonstrate that incorporating spatial context into service composition is an NP-Hard problem and model it as an integer linear program. We propose two heuristic approaches capable of producing near-optimal solutions in real-time. We implement a simulation of the composition problem to study the performance of our approaches. Our experimental results show that the proposed methods are scalable compared to the branch-and-cut algorithm. These results set a precedent against which future work on solutions to this problem can be compared.

Network slicing is one of the major catalysts to turn future telecommunication networks into versatile service platforms. Along with its benefits, network slicing is introducing new challenges in the development of sustainable network operations. In fact, guaranteeing slices requirements comes at the cost of additional energy consumption, in comparison to non-sliced networks. Yet, one of the main goals of operators is to offer the diverse 5G and beyond services, while ensuring energy efficiency. To this end, we study the problem of slice activation/deactivation, with the objective of minimizing energy consumption and maximizing the users quality of service (QoS). To solve the problem, we rely on two Multi-Armed Bandit (MAB) agents to derive decisions at individual base stations. Our evaluations are conducted using a real-world traffic dataset collected over an operational network in a medium size French city. Numerical results reveal that our proposed solutions provide approximately 11-14% energy efficiency improvement compared to a configuration where all the slice instances are active, while maintaining the same level of QoS. Moreover, our work explicitly shows the impact of prioritizing the energy over QoS, and vice versa.

Federated Learning (FL) is a distributed machine learning technique where multiple devices can collaboratively train a model without sharing their data. As a result, FL ensures distinct privacy benefits compared to centralized training approaches. However, despite its benefits, FL remains susceptible to reverse-engineering attacks that can uncover sensitive information about the training data from the local updates sent by each participant. To address this issue, we propose a framework for securely and efficiently aggregating the results of FL on multiple devices. We compare and evaluate the performance of two techniques, Homomorphic Encryption (HE) and Secure Multiparty Computation (SMPC), to determine the best method that allows devices to share their learning results without revealing their raw local data. Ultimately, we propose using SMPC protocol as the most effective solution to secure FL. Our framework is experimentally evaluated, and its effectiveness in terms of security, efficiency, and accuracy is demonstrated.

Sixth-generation and mobile edge computing (MEC) systems are expected to empower a wide range of applications. Unmanned aerial vehicles (UAVs) can play a vital role in improving network connectivity. Hence, our problem is to minimize the user equipment (UE) energy consumption during task offloading in a UAV assisted MEC system. To address the formulated NP-hard problem, we propose task scheduling and assignment algorithms for mapping UE tasks to fixed edge servers using UAV. Lastly, the simulation results demonstrate that the proposed algorithms yield better results than other benchmark methods in terms of total UE energy consumption.

Next-generation mobile networks are expected to meet the requirements of a wide range of new vertical services. Hence, the network slicing concept has been introduced, in which Mobile Virtual Network Operators (MVNOs) are allowed to provide various types of services over the same physical infrastructure, owned by an Infrastructure Provider (InP). To cope with an ever-changing traffic demand, MVNOs seek to pre-allocate/reconfigure the resources at the base stations in an anticipatory manner, based on traffic demand predictions. Ideally, conducting per-slice traffic forecasting requires information that is likely to disclose MVNO confidential information (i.e., business strategy or private user data). To secure data ownership while conducting traffic forecasting, we propose the Federated Proximal Long Short-Term Memory (FPLSTM) framework, which allows MVNOs to train their local models with their private dataset at each base station; subsequently, an associated InP global model can be updated through the aggregation of the local models. The results obtained by training the models on a real-world dataset indicate that the forecasting performance of our proposed approach is as accurate as state-of-the-art centralized solutions, while improving data privacy. To enable scalability, we further propose the Information-based Clustering FPLSTM (IC-FPLSTM) and Random Clustering FPLSTM (RC-FPLSTM) frameworks, dealing with large-scale cellular networks. These solutions demonstrate computation and communication cost efficiency significantly above the state-of-the-art.

In this work, taking the perspective of Mobile Virtual Network Operators (MVNOs), we tackle the multi-slice traffic forecasting problem, while respecting the data privacy of users. To this end, we propose the Federated Proximal Long Short-Term Memory (FPLSTM) framework, which allows MVNOs to train at each base station their local models with their private datasets, without compromising data privacy. Prediction results obtained by evaluating the models on a real-world dataset indicate that the forecast of FPLSTM is as accurate as state-of-the-art solutions while ensuring data privacy as well as computation and communication costs efficiency.

Network slicing is one of the cornerstones for next-generation mobile communication systems. Specifically, it enables Mobile Virtual Network Operators (MVNOs) to offer various types of services over the same physical infrastructure owned by an Infrastructure Provider (InP). To satisfy the dynamic user requirements and ensure resource efficiency, MVNOs need to estimate the future traffic demand in advance, to pre-allocate/reconfigure the resources at the base stations. However, this per-slice traffic forecasting exploits information that is clearly sensitive for the MVNOs from a business point of view, and which might even disclose private data regarding some users. Hence, it is vital for MVNOs to ensure data privacy while conducting traffic forecasting. Bearing this in mind, we propose the Federated Proximal Long Short-Term Memory (FPLSTM) framework, which allows MVNOs to train their local models with their private dataset at each base station without compromising data privacy. Simultaneously, an InP global model is updated through the aggregation of local models weights. Prediction results obtained by training the models on a real-world dataset indicate that the forecasting performance of FPLSTM is as accurate as state-of-the-art solutions, while ensuring data privacy, computation and communication cost efficiency.

The exponential growth of traffic in mobile networks is leading to increased pressure on the infrastructure of mobile networks and in particular on their backhaul networks. It is more critical than ever to carefully plan backhaul networks. In this paper, we formulate and solve the problem of hierarchical wireless backhaul network design. In formulating our problem, we cover different requirements, namely: topology simplicity, network resiliency, and link reliability. We formulate the problem as an Integer Linear Programming (ILP) problem, allowing us to solve the problem to optimality. Furthermore, we provide a graph theory-based algorithm that allows solving the problem over a large scale. The proposed algorithm exploits the properties of the graph that represents the network. The results of our evaluations in various network scenarios demonstrate the efficiency of our ILP formulation and the proposed algorithm in keeping the backhaul network simple, resilient, and reliable. Using a practical channel propagation model and different node densities that are representative of small-scale and large-scale urban environments, our results also show that even with high resiliency requirements, the network traffic can be backhauled with only 5-10% of the nodes for the considered densities. Our results also demonstrate that our algorithm leads to near-optimal solutions in different scenarios.

Users' viewing experience in the video delivery process is of paramount importance for Content Delivery Networks (CDNs). Throughout their operations, CDN providers target the satisfaction of users' expectations in terms of Quality of Experience (QoE). In this context, CDN providers need to acquire knowledge on users' QoE and correlate observations through different video sessions in order to identify QoE degradations and investigate their potential root cause. In the absence of users' feedback on their QoE, CDN providers can monitor and analyze Key Performance Indicators (KPIs) throughout video sessions. This allows to assess the Quality of Service (QoS) offered to users, influencing their QoE. However, due to the large number of sessions handled by CDN operators, it is not possible to conduct such an analysis manually. In this work, we introduce a framework that allows to automatically group a large set of video sessions into a small number of representative clusters, with each cluster containing video sessions with similar patterns of KPIs. The framework builds upon a set of features representing the evolution of KPIs over a session. It relies on an unsupervised machine learning algorithm to form the clusters. We evaluate the framework over a real-world dataset with traffic logs relating to thousands of sessions. The obtained results underline the capabilities of the proposed framework.

Multimedia conferencing is the conversational exchange of multimedia content between multiple parties. It has a wide range of applications (e.g., massively multiplayer online games (MMOGs) and distance learning). Media handling services (e.g., video mixing, transcoding, and compressing) are critical to multimedia conferencing. However, efficient resource usage and scalability still remain important challenges. Unfortunately, the cloud-based approaches proposed so far have several deficiencies in terms of efficiency in resource usage and scaling, while meeting quality of service (QoS) requirements. This paper proposes a solution which optimizes resource allocation and scales in terms of the number of participants while guaranteeing QoS. Moreover, our solution composes different media handling services to support the participants' demands. We formulate the resource allocation problem mathematically as an integer linear programming (ILP) problem and design a heuristic for it. We evaluate our proposed solution for different numbers of participants and different participants' geographical distributions. Simulation results show that our resource allocation mechanism can compose the media handling services and allocate the required resources in an optimal manner while honoring the QoS in terms of end-to-end delay.

Multimedia conferencing is used extensively in a wide range of applications, such as online games and distance learning. These applications need to efficiently scale the conference size as the number of participants fluctuates. Cloud is a technology that addresses the scalability issue. However, the proposed cloud-based solutions have several shortcomings in considering the future demand of applications while meeting both Quality of Service (QoS) requirements and efficiency in resource usage. In this paper, we propose an Adaptive and Dynamic Scaling mechanism (ADS) for multimedia conferencing services in the cloud. This mechanism enables scalable and elastic resource allocation with respect to the number of participants. ADS produces a cost efficient scaling schedule while considering the QoS requirements and the future demand of the conferencing service. We formulate the problem using Integer Linear Programming (ILP) and design a heuristic for it. Simulation results show that ADS mechanism elastically scales conferencing services. Moreover, the ADS heuristic is shown to outperform a greedy algorithm from a resource-efficiency perspective.

The development of virtualization techniques enables an architectural shift in mobile networks, where resource allocation, or even signal processing, become software functions hosted in a data center. The centralization of computing resources and the dynamic mapping between baseband processing units (BBUs) and remote antennas (RRHs) provide an increased flexibility to mobile operators, with important reductions of operational costs. Most research efforts on Cloud Radio Access Networks (CRAN) consider indeed an operator perspective and network-side performance indicators. The impact of such new paradigms on user experience has been instead overlooked. In this paper, we shift the viewpoint, and show that the dynamic assignment of computing resources enabled by CRAN generates a new class of mobile terminal handover that can impair user quality of service. We then propose an algorithm that mitigates the problem, by optimizing the mapping between BBUs and RRHs on a time-varying graph representation of the system. Furthermore, we show that a practical online BBU-RRH mapping algorithm achieves results similar to an oracle-based scheme with perfect knowledge of future traffic demand. We test our algorithms with two large-scale real-world datasets, where the total number of handovers, compared with the current architectures, is reduced by more than 20%. Moreover, if a small tolerance to dropped calls is allowed, 30% less handovers can be obtained.

Wireless Sensor Networks (WSNs) are being used extensively today in various domains. However, they are traditionally deployed with applications embedded in them which precludes their re-use for new applications. Nowadays, virtualization enables several applications on a same WSN by abstracting the physical resources (i.e. sensing capabilities) into logical ones. However, this comes at a cost, including an energy cost. It is therefore critical to ensure the efficient allocation of these resources. In this paper, we study the problem of assigning application sensing tasks to sensor devices, in virtualized WSNs. Our goal is to minimize the overall energy consumption resulting from the assignment. We focus on the static version of the problem and formulate it using Integer Linear Programming (ILP), while accounting for sensor nodes' available energy and virtualization overhead. We solve the problem over different scenarios and compare the obtained solution to the case of a traditional WSN, i.e. one with no support for virtualization. Our results show that significant energy can be saved when tasks are appropriately assigned in a WSN that supports virtualization.

Traditional Content Delivery Networks (CDNs) built with traditional Internet technology are less and less able to cope with today's tremendous growth of content. Information Centric Networks (ICN), a proposed future Internet technology, may aid in remedying the situation. Unlike the current Internet, it decouples information from its sources and provides in- network storage. We expect traditional CDN and ICN-based CDN to co-exist in the foreseeable future, especially as it is now known that it might be possible to evolve traditional CDNs to gain the benefits promised by ICN. 5G providers must therefore aim to offer core network slices on which both ICN-based CDNs and traditional CDNs can be built. These slices could of course also be offered to providers of other applications with requirements similar to those of content delivery. This paper tackles the problem of slicing 5G for content delivery over ICN- based CDNs and traditional CDNs. Only virtualized Evolved Packet Core (EPC)-based 5G is considered. The problem is defined as a resource allocation problem which aims at minimizing the cost of slice assignment, while meeting QoS requirements. An Integer linear programming (ILP) formulation is provided and evaluated in a small-scale scenario.

The European Telecommunications Standards Institute (ETSI) developed Network Functions Virtualization (NFV) Management and Orchestration (MANO) framework. Within that framework, NFV orchestrator (NFVO) and Virtualized Network Function (VNF) Manager (VNFM) functional blocks are responsible for managing the lifecycle of network services and their associated VNFs. However, they face significant scalability and performance challenges in large-scale and geo-distributed NFV systems. Their number and location have major implications for the number of VNFs that can be accommodated and also for the overall system performance. NFVO and VNFM placement is therefore a key challenge due to its potential impact on the system scalability and performance. In this paper, we address the placement of NFVO and VNFM in large-scale and geo-distributed NFV infrastructure. We provide an integer linear programming formulation of the problem and propose a two-step placement algorithm to solve it. We also conduct a set of experiments to evaluate the proposed algorithm.

In the next few years, mobile networks will undergo significant evolutions in order to accommodate the ever-growing load generated by increasingly pervasive smartphones and connected objects. Among those evolutions, cognitive networking upholds a more dynamic management of network resources that adapts to the significant spatiotemporal fluctuations of the mobile demand. Cognitive networking techniques root in the capability of mining large amounts of mobile traffic data collected in the network, so as to understand the current resource utilization in an automated manner. In this paper, we take a first step towards cellular cognitive networks by proposing a framework that analyzes mobile operator data, builds profiles of the typical demand, and identifies unusual situations in network-wide usages. We evaluate our framework on two real-world mobile traffic datasets, and show how it extracts from these a limited number of meaningful mobile demand profiles. In addition, the proposed framework singles out a large number of outlying behaviors in both case studies, which are mapped to social events or technical issues in the network.

Understanding of the network topology is a basic building block towards the design of efficient networking solutions. In the context of vehicular networks, such a step is especially crucial due to the highly dynamic nature of vehicles that can lead to strong instantaneous variations in the structure of the network. This notwithstanding, and despite the soon-to-come real-world deployment of vehicle-to-vehicle communication technologies, we still lack a clear understanding of vehicular network topological properties. In this paper, we present a complex network analysis of the instantaneous topology of a realistic vehicular network in Cologne, Germany. Our study unveils a poorly connected topology, with very limited availability, reliability, and navigability. We also examine the vehicular network topology in a second scenario, i.e., Zurich, Switzerland. The comparative analysis shows how simplistic mobility models can lead to unrealistic overly connected topologies.

The European Telecommunications Standards Institute management and orchestration framework is designed to enable automated management of the network services and their corresponding virtualized network functions (VNFs). In that context, the network function virtualization (NFV) orchestrator (NFVO) manages the lifecycle of the network services and coordinates with the VNF managers (VNFMs) which manage the VNFs lifecycle. In large-scale and distributed NFV deployments, these management functions face critical challenges such as delays, and variations in VNFs workload. Placing NFVO and VNFM in a large-scale distributed NFV deployment is therefore a very challenging problem due to the potential negative impact on performance and operational cost. However, to the best of our knowledge, the problem has not yet been addressed in the literature. This paper is a first step toward a solution to the overall problem. It focuses on the VNFM placement and aims at minimizing the operational cost without violating the performance requirements. We call this the VNFM placement problem (MPP). We provide an integer linear programming formulation and propose a tabu search algorithm to solve larger instances of the MPP. We evaluate our algorithm in a real-world large-scale scenario. The results show that it leads to significant reductions in the operational cost of large-scale distributed NFV deployments.

Cellular communications are undergoing significant evolutions in order to accommodate the load generated by increasingly pervasive smart mobile devices. Dynamic access network adaptation to customers' demands is one of the most promising paths taken by network operators. To that end, one must be able to process large amount of mobile traffic data and outline the network utilization in an automated manner. In this paper, we propose a framework to analyze broad sets of Call Detail Records (CDRs) so as to define categories of mobile call profiles and classify network usages accordingly. We evaluate our framework on a CDR dataset including more than 300 million calls recorded in an urban area over 5 months. We show how our approach allows to classify similar network usage profiles and to tell apart normal and outlying call behaviors.

There is a growing need for vehicular mobility datasets that can be employed in the simulative evaluation of protocols and architectures designed for upcoming vehicular networks. Such datasets should be realistic, publicly available, and heterogeneous, i.e., they should capture varied traffic conditions. In this paper, we contribute to the ongoing effort to define such mobility scenarios by introducing a novel set of traces for vehicular network simulation. Our traces are derived from high-resolution real-world traffic counts, and describe the road traffic on two highways around Madrid, Spain, at several hours of different working days. We provide a thorough discussion of the real-world data underlying our study, and of the synthetic trace generation process. Finally, we assess the potential impact of our dataset on networking studies, by characterizing the connectivity of vehicular networks built on the different traces. Our results underscore the dramatic impact that relatively small communication range variations have on the network. Also, they unveil previously unknown temporal dynamics of the topology of highway vehicular networks, and identify their causes.

Despite the growing interest in a real-world deployment of vehicle- to-vehicle communication, many topological features of the resulting vehicular network remain largely unknown. We still lack a clear understanding of the level of connectivity achievable in large-scale urban scenarios, of the availability and reliability of connected multi-hop paths, and of the evolution of such features over daytime. In this paper, we investigate how the instantaneous topology of the vehicular network would look like in the case of Cologne, Germany, a typical middle-sized European city. Through a complex network analysis, we unveil the low connectivity, availability, reliability and navigability of the network, and exploit our findings to derive network design and usage guidelines.