Search

With the growth in the use of smart mobile devices and the development of information technologies, worms and malware can spread from mobile networks into heterogeneous and hierarchical networks. Thus, the spread of these worms constitutes an increasing potential threat. For understanding the propagation of the aforementioned wireless hybrid worms, current researches have three critical problems: Structural simplification of network topologies (previous research object for wireless worms is the mobile network), Homogenous population of network devices (properties of network devices are the same), and Inaccuracy of propagation models (traditional deterministic differential or stochastic difference models cannot model propagation of wireless hybrid worms accurately). To address them, we propose a novel compartmental population-based propagation model oriented towards heterogeneous and hierarchical networks with human behaviors, and then study the impacts of user mobility and operation behaviors on worm propagation. Meanwhile, we conduct extensive simulations to show our model can characterize propagation features accurately. The results in this paper not only provide a deep understanding of new worm propagation, but also serve as fundamental defense guidelines.

Nowadays, with the explosive growth of sensor-based devices connected to Internet of Thing (IoT), massive amount of data are generated every day with potential tremendous value. We argue that the value of those data can be extracted through monetize data platform in IoT-Edge-Cloud ecosystems for many parts of the business. In such monetize data platform, the data can be computed and transformed into services in IoT-Edge-Cloud ecosystems and provide Data-As-A-Service (DAAS) for applications. The key to implement such a monetize data platform is to evenly distribute DAAS computing tasks to network devices to maximize the benefits of the system. So, in this paper, we study the Task Type-based Computation Offloading algorithm (TTCO) to implement such platform. We use the "IoT-Edge-Cloud" three-layer multi-hop model, which is closer to the complex scene in monetize data platform. We divide tasks into data-intensive tasks and CPU-intensive tasks, and then combine the cost model of computation offloading with task type to make data-intensive tasks prefer local computing and CPU-intensive tasks prefer offload computing, thereby reducing the monetize data platform transmission volume and improving the overall quality of computation offloading. We then use a hierarchical game model combined with fictitious play to solve the Nash Equilibrium (NE) of the system and obtain the mixed strategies of the devices. Finally, we propose a TTL-constrained flood strategy transmission mechanism to make the algorithm apply to practice. The experimental results prove that our algorithm has a large performance gain in various scenarios, which can be severed as a monetize data platform for IoT-Edge-Cloud ecosystems.

Despite the advantages of multi-access edge computing in enabling latency-sensitive services and extending the limited computing capabilities of network devices, access communication issues are still often causing the quality of the wireless channels to be severely degraded, preventing the edge resources from being efficiently utilized. Through the deployment of low-cost passive reflecting elements, the recent studies of intelligent reflecting surfaces (IRSs) in wireless networks have shown a great potential for enhancing the quality of the wireless channels and the transmission rates. In this work, motivated by the recent findings, we study the use of an IRS-aided edge computing system for enabling low latency and high reliability computation offloading in the context of a single-user network. Specifically, we optimize the phase shift of the IRS elements along with the device’s transmit power and offloading decision, with the objective of minimizing the device’s energy consumption. Due to the non-convexity of the problem, we propose a customized sub-optimal solution based on the alternating optimization approach, utilizing novel successive convex approximation techniques. Numerical analysis demonstrates the energy reduction and saving in network resources provided by the optimized use of the IRS, especially for offloading services with higher reliability.

Cloud computing applications have grown rapidly in the last decade, where many organizations are migrating their applications to cloud data center as they expected high performance, reliability, and the best quality of service. Data centers deploy a variety of technologies, such as software-defined networks (SDN), to effectively manage their resources. The SDN approach is a highly flexible network architecture that automates network configuration using a centralized controller to overcome traditional network limitations. This paper proposes an SDN-based monitoring algorithm to detect the performance anomaly and identify the bottleneck of the distributed application in the cloud data center using the support vector machine algorithm. It collects the data from the network devices and calculates the performance metrics for the distributed application components that are used to train the SVM algorithm and build a baseline model of the normal behavior of the distributed application. The SVM model detects performance anomaly behavior and identifies the root cause of bottlenecks using one-class support vector machine (OCSVM) and multi-class support vector machine (MCSVM) algorithms. The proposed method does not require any knowledge about the running applications or depends on static threshold values for performance measurements. Simulation results show that the proposed method can detect and locate the failure occurrences efficiently with high precision and low overhead compared to statistical methods, Naive Bayes Classifier and the decision tree machine learning method.

The diverse and stringent requirements of 6G networks, detailed in the present book, give rise to fully flexible, end‐to‐end, software‐defined network paradigm, where every part of the network can be configured and controlled via software. However, the wireless environment per se – i.e. the wireless medium or channel – is generally assumed uncontrollable, thus, imposing inherent limitations to 6G networks, that arise by the very nature of wireless operation. To this end, the emergence of the newly introduced concept of reconfigurable intelligent surfaces (RISs) challenges the fundamental status quo that the wireless environment is fixed by nature, paving the way for the realization of end‐to‐end fully reconfigurable 6G networks.RISs are network devices that can control and manipulate the radio waves traversing through the wireless channel. Since RISs reside within the wireless channel and not at the communication endpoints, they can control the wireless channels from within. In this context, the present chapter introduces the concept of RISs within the 6G framework, elaborating on their advantages and limitations. RISs are then compared with other transmission technologies, e.g. phased arrays, multi‐antenna transmitters, and relays, while demonstrating the trade‐offs governing their operation and applicability. Subsequently, the different types of RIS implementations are presented from a theoretical standpoint, along the available prototypes in the literature; thus, demonstrating how the RIS technology is already a reality, ushering wireless networks into the 6G era.

Cross-border trading platforms study logistics from the perspective of the network, involving all links in the network and the supply chain. The network status is constantly changing, and the acquisition and analysis of the real-time status of each link on the transaction path can further improve the accuracy of the optimal path selection. Meanwhile, traditional networks can no longer adapt to the rapid development of today’s network technology and applications due to their rigid structure and simplified functions. As for traditional networks and their associated network equipment, problems such as inconvenient management and monitoring, poor interoperability, difficulty in expanding network functions, and high operating costs have severely hindered the efficiency of network functions and the pace of innovation. In order to solve many problems faced by traditional networks, in the paper, an SDN network path selection algorithm based on the dual impact factors of cloud service nodes is proposed. In the algorithm proposed in the paper, the final selected path is considered to be composed of multiple links. First, a cloud environment based on the Ryu controller is build, and a corresponding data collection strategy is constructed to complete the acquisition of the link available bandwidth and delay that affect the real-time status of the path. Then, the weighted results are calculated according to the proposed candidate path weight distribution formula to realize the sorting and screening of each transaction link, and provide support for the selection of the cross-border transaction system service path with load balance. The simulation results show that compared with the traditional DLB algorithm, the algorithm proposed in the paper has better load balancing performance, which further shows that the proposed algorithm has a better choice for the path.

Network programmers can currently deploy an arbitrary set of protocols in forwarding devices through data plane programming languages such as P4. However, as any other type of software, P4 programs are subject to bugs and misconfigurations. Network verification tools have been proposed as a means of ensuring that the network behaves as expected, but these tools frequently face severe scalability issues. In this paper, we argue for a novel approach to this problem. Rather than statically inspecting a network configuration looking for bugs, we propose to enforce networking properties at runtime. To this end, we developed P4box , a system for deploying runtime monitors in programmable data planes. P4box allows programmers to easily express a broad range of properties (both program-specific and network-wide). Moreover, we provide an automated framework based on assertions and symbolic execution for ensuring monitor correctness. Our experiments on a SmartNIC show that P4box monitors represent a small overhead to network devices in terms of latency, throughput and power consumption.

Virtual network embedding (VNE) in network virtualization over flexi-grid networks is one of the main challenges up to now. Many models and algorithms based on reducing the cost or keeping integrity and continuity of spectrum slots have been already proposed. In this paper, we employ the concept of distribution equilibrium of computation resources to improve performance of VNE. An integer linear programming (ILP) model and an algorithm based on the concept of distribution equilibrium for virtual node embedding are developed. The experimental results show that this algorithm has some significant advantages. And then, an ILP model based on resource distribution equilibrium and adaptive modulation for VNE over flexi-grid optical networks is established. Since the ILP problem is NP-complete, a heuristic algorithm to manipulate VNE on table is developed. The simulation results reveal the superiority in terms of acceptance ratios and system revenues.

In this paper, we propose a versatile network architecture, named VFTree, for data centers. VFTree provisions unique versatility in the sense that its topology and performance can be tailored with great freedom. The wide dynamic range and fine granularity of its configurations allow this architecture to closely follow the demands of the running traffic. Also, this versatility facilitates the reuse of network devices, so that upgrades could be achieved by rearranging existing devices, avoiding the expenditure on new generations of hardware in each upgrade cycle. Furthermore, this versatility allows VFTree to be elaborately tuned into cost efficient zones. This way, compared to widely adopted industrial solutions including fat-tree, VFTree can provide even higher aggregate bandwidth and richer path availability at the same cost. In this paper, we first design the network topology of VFTree. We parameterize the topology so that it is flexible and generalized. For example, we can configure the numbers of upward and downward ports for the switches, which are fixed in the classic fat-tree. Based on that, we propose routing algorithms, implement the abstraction from packet level to the flow level, and handle the flow collisions. We also analyze VFTree’s performance, demonstrate its aforementioned features, and compare it with its predecessors. VFTree can be seen as a generalization of, and is backward-compatible to fat-tree. We believe it is ready to be deployed in industry, as a perfect replacement of the fat-tree DCNs.

Traditional telecommunication networks were designed for the use of specialized hardware devices (routers, Ethernet switches, etc.). These devices were created on the basis of specific hardware and software platforms of individual vendors. The deployment of such network devices has led to long cycles of design and commissioning, and, consequently, to a slowdown in the introduction of new products and services. Maintenance and management of networks of this type has been and remains quite inefficient and expensive. The sharing of SDN and NFV changes the traditional paradigm of network construction, which is how the operator designs, develops, administers the network, and provides products and services to users. Such a paradigm shift can provide many technological and operational benefits. The paradigm shift is aimed at a fundamental rethinking of the cost structure of the operator and the mode of its operational processes. This shift, when used properly, is also able to provide fast and flexible on-demand services development, which increases the operator's competitiveness in the telecommunications and information services market. The article analyzes the advantages of implementing modern technologies for virtualization of network functions. Combining SDNs with NFVs has great advantages, but the problem is their integration. In general, the article is devoted to the study of ways to integrate virtualized network functions and the SDN controller. There are two possible architectures for this integration: the SDN controller interacts with virtualized network functions (VNF) or the switch interacts with VNF. This article provides an analytical description of both options. Thus, the article aims to create a mathematical model for analytical modeling of SDN with NFV and ultimately increase productivity and accelerate the introduction of modern network technologies.

Proliferation of large number of cloud users steered the exponential increase in number and size of the data centers. These data centers are energy hungry and put burden for cloud service provider in terms of electricity bills. There is environmental concern too, due to large carbon foot print. A lot of work has been done on reducing the energy requirement of data centers using optimal use of CPUs. Virtualization has been used as the core technology for optimal use of computing resources using VM migration. However, networking devices also contribute significantly to the responsible for the energy dissipation. We have proposed a two level energy optimization method for the data center to reduce energy consumption by keeping SLA. VM migration has been performed for optimal use of physical machines as well as switches used to connect physical machines in data center. Results of experiments conducted in CloudSim on PlanetLab data confirm superiority of the proposed method over existing methods using only single level optimization.

Software-Defined Vehicular Networks (SDVNs) have been a vital component to various radio access technologies that supports massive data loads in various security and infotainment applications. SDVNs elevate the constraints of static hardware network devices to programmable units and provide a global view of the network's status and a common interface between varied radio access technologies. However, having a logically centralized control unit comes with several challenges, including bottleneck issues and densification issues. The distributed control plane appears as a possible alternative to the centralized one yet raises questions concerning where to deploy control units, how many SDN controllers are needed in a given network structure, and when to migrate switches or change controllers. Therefore, the development of an efficient topology management entity is an important aspect that needs to be addressed. In this article, we define the main components necessary to develop and design an efficient SDVN topology management architecture. We evaluate the current solutions and methods of controllers' deployment and assignment for software-defined vehicle networks. We then discuss several challenges, design guidelines, open issues, and future directions for research related to the development of future adaptive software-defined vehicular networks.

The past few years have witnessed the compelling applications of the Internet of Things (IoT) in our daily life. Meanwhile, with the explosion of IoT devices and various applications, the expectations for the performance, reliability, and security of networks are greater than ever. The current end-host-based or centralized control framework incurs too much communication and computation overhead, therefore exhibiting tardiness and clumsiness in responding to network dynamics. Recently, with the advancement of programmable network hardware, it is possible to implement network functions inside the network. However, current in-network schemes are largely dependent on the manual process, which presents poor scalability and robustness. Therefore, in this article, we present a new intelligent network control architecture, in-network intelligence control. We design intelligent in-network devices that can automatically adapt to network dynamics by leveraging powerful machine learning adaptive abilities. In addition, to enhance the collaboration among distributed in-network devices, a centralized management plane is introduced to ease the training process of distributed switches. To demonstrate the technical feasibility and performance advantage of our architecture, we present three use cases: in-network load balance, in-network congestion control, and in-network DDoS detection.

Recently, smart cities, healthcare system, and smart vehicles have raised challenges on the capability and connectivity of state-of-the-art Internet-of-Things (IoT) devices, especially for the devices in hotspots area. Multiaccess edge computing (MEC) can enhance the ability of emerging resource-intensive IoT applications and has attracted much attention. However, due to the time-varying network environments, as well as the heterogeneous resources of network devices, it is hard to achieve stable, reliable, and real-time interactions between edge devices and their serving edge servers, especially in the 5G ultradense network (UDN) scenarios. Ultradense edge computing (UDEC) has the potential to fill this gap, especially in the 5G era, but it still faces challenges in its current solutions, such as the lack of: 1) efficient utilization of multiple 5G resources (e.g., computation, communication, storage, and service resources); 2) low overhead offloading decision making and resource allocation strategies; and 3) privacy and security protection schemes. Thus, we first propose an intelligent UDEC (I-UDEC) framework, which integrates blockchain and artificial intelligence (AI) into 5G UDEC networks. Then, in order to achieve real-time and low overhead computation offloading decisions and resource allocation strategies, we design a novel two-timescale deep reinforcement learning (2Ts-DRL) approach, consisting of a fast-timescale and a slow-timescale learning process, respectively. The primary objective is to minimize the total offloading delay and network resource usage by jointly optimizing computation offloading, resource allocation, and service caching placement. We also leverage federated learning (FL) to train the 2Ts-DRL model in a distributed manner, aiming to protect the edge devices’ data privacy. Simulation results corroborate the effectiveness of both the 2Ts-DRL and FL in the I-UDEC framework and prove that our proposed algorithm can reduce task execution time up to 31.87%.

The next decade will usher in a new era of technological innovation where compute, communications, and intelligence will converge. The number of connected devices is estimated to reach 500 billion by 2030, which is 59× larger than the expected world population [1] . Instead of humans dominating the use of wireless networks, objects or things will become the dominant users. Networks will incorporate intelligent compute to enable new types of applications such as cyberphysical worlds that replicate reality. These applications will require distributed intelligent compute platforms, as processing terabytes of data at a single edge device will be near impossible. Future 6G systems are currently being defined to meet these needs, as shown in Figure 1 , and the essential components of all 6G systems will be intelligence and security [2] . These new wireless systems will need to leverage a distributed compute network that is part of the wireless network so that the edge device can employ a high-performance compute platform that is in close proximity to itself. Such a network device must incorporate flexibility and programmability at low latency while also supporting network and connectivity functions. Heterogeneous integration in 2.5D and 3D can efficiently enhance existing compute platforms such as processors, application-specific integrated circuits, and field programmable gate arrays (FPGAs) to address emerging applications such as intelligent compute-communicate systems at lower cost with faster time to market. This article presents an overview of 2.5D and 3D heterogeneous technologies and how they can be leveraged to create new intelligent-compute-communicate platforms to enable emerging applications

As networks grow larger in size and complexity, their monitoring is becoming an increasing challenge because of the required polling performance and also due to heterogeneity of devices. As it turns out, SNMP (Simple Network Management Protocol) is by far the most popular monitoring protocol. However, due to the increase in the number of network devices, it becomes necessary to employ multiple SNMP managers, which is not cost-effective due to the hardware requirements. Additionally, the different proprietary SNMP implementations require custom configuration very often, as new devices are being incorporated into the network. Therefore, current SNMP managers not only require capabilities for large-scale monitoring but also a high degree of flexibility and programmability. In response, we propose an SNMP manager architecture with a flexible multi-threaded architecture, which effectively reduces the hardware resources necessary to poll the increasing number of SNMP agents. In addition, it features a scripting component to deal with the different data representations caused by proprietary implementations. Our experience has shown that SNMP agents can have high variability in their response times. Actually, our findings show a strong correlation between high response times and CPU load. As a solution, we propose and analyze novel adaptive polling algorithms that decrease the load on agents' CPUs while keeping the desired polling rate for fast agents. Finally, we present several real-world use cases where we show the benefits of the polling algorithms and the scripting component, by means of extensive measurement campaigns, This work was supported by Ayudas para la formación de doctores en empresas, Doctorados Industriales, under Grant DI-16-08979

International audience; In order to improve locality aspects, new Cloud-related architectures such as Edge Computing have been proposed. Despite the growing popularity of these new architectures, their energy consumption has not been well investigated yet. To move forward on such a critical question, we first introduce a taxonomy of different Cloud-related architectures. From this taxonomy, we then present an energy model to evaluate their consumption. Unlike previous proposals, our model comprises the full energy consumption of the computing facilities, including cooling systems, and the energy consumption of network devices linking end users to Cloud resources. Finally, we instantiate our model on different Cloud-related architectures, ranging from fully centralized to completely distributed ones, and compare their energy consumption. The results show that a completely distributed architecture, because of not using intra-data center network and large-size cooling systems, consumes between 14% and 25% less energy than fully centralized and partly distributed architectures respectively. To the best of our knowledge, our work is the first one to propose a model that enables researchers to analyze and compare energy consumption of different Cloud-related architectures.

Large-scale datacenters are the key infrastructures of cloud computing. Inside a datacenter, a large number of servers are interconnected using a specific datacenter network to deliver the infrastructure as a service (IaaS) for tenants. To realize novel cloud applications like the network virtualization and network isolation among tenants, the principle of software-defined network (SDN) has been applied to datacenters. In the setting, multiple distributed controllers are deployed to offer a control plane over the entire datacenter to efficiently manage the network usage. Despite such efforts, cloud datacenters, however, still lack a scalable and resilient control plane. Consequently, this paper systematically studies the coverage problem of controllers, which means to cover all network devices using the least number of controllers. More precisely, we tackle this essential problem from three aspects, including the minimal coverage, the minimal fault-tolerant coverage, and the minimal communication overhead among controllers. After modelling and analyzing such three problems, we design efficient approaches to approximate the optimal solution, respectively. Extensive evaluation results indicate that our approaches can significantly save the number of required controllers, improve the fault-tolerant capability of the control plane and reduce the communication overhead of state synchronization among controllers. The design methodologies proposed in this paper can be applied to cloud datacenters with other networking structures after minimal modifications.

In the context of wide area networks (WANs) and software defined networking (SDN), reducing the communication delays experienced by the network devices is an important challenge whose solution requires a careful placement of controllers to decrease the end-to-end latencies. Although the majority of studies focusing on the controller placement problem (CPP) only consider the switch–controller propagation delays and inter-controller latencies, the controllers’ capacity needs to be addressed for lowering the end-to-end delays. While the controllers with a variety of processing rates, number of ports, and costs are available in the market, Internet Service Providers (ISPs) need to consider the affordability and capability of compensating their needs by maximizing the use of their networks’ resources. To propose a solution for this important problem, we consider the Knapsack 0–1problem and formulate the Garter Snake Optimization Capacitated Controller Placement Problem (GSOCCPP), a meta-heuristic algorithm, with new iterations and temperate mating conditions to solve the CPP. This algorithm uses a reasonable amount of computation time to obtain the minimum delays. A number of Topology-Zoo datasets are analyzed with a variety of small to large scale data plane nodes to evaluate the GSOCCPP algorithm based on different controllers’ capacities. The simulation results demonstrate that, in addition to outperforming similar meta-heuristic and clustering algorithms such as the Firefly Algorithm, Particle Swarm Optimization, and the k-means++, our newly proposed GSOCCPP algorithm is successful in achieving the lowest execution time among the analyzed algorithms. Furthermore, this proposed solution has a more efficient memory consumption compared to other algorithms for controller placement in different network topologies.

Self-driving networks represent the next step of network management techniques in the close future. A fundamental point for such an evolution is the use of Machine Learning based solutions to extract information from data coming from network devices during their activity. In this work we focus on a new type of data, available thanks to the definition of the novel SRv6 paradigm, referred to as SRv6 Traffic Counters ( SRTCs ). SRTCs provide aggregated measurements related to forwarding operations performed by SRv6 routers. In this work a detailed description of different SRTCs types (SR.INT, PISD, PSID.TM and POL) is provided and their relationships is formalized. The theoretical framework deployed is used to identify, on the basis of network configuration parameters of both SRv6 and IGP protocols, the minimum set of independent SRTCs to characterize the Network Status : we show that about the 80% of counters can be neglected with no information loss. We also apply our framework to two use cases: i) Traffic Matrix (TM) Assessment and ii) Traffic Anomaly Detection. For the TM assessment, we show that in a partially deployed SRv6 scenario a specific type of SRTCs , i.e., PSID, is more reliable than other ones; on the contrary, in a fully deployed scenario POL and PSID.TM counters provide the full TM knowledge. For the Traffic Anomaly Detection case, we show that known solutions based on link load measurements can be improved when integrating SRTCs information.

The unprecedented growth in data volume results in an urgent need for a dramatic increase in the size of data center networks. Accommodating millions to even billions of servers in the structure of data center networks, while providing highly-efficient interconnecting architectures, is of prominent importance nowadays. Generally, high scalability, noticeable fault-tolerance, great network capacity, limited delay, reasonable power consumption, and justifiable construction cost have to be listed as the principal design goals of well-organized cloud data center networks. In this paper, Platonica, a novel dual-centric and recursively defined architecture, is proposed. Platonica is constructed by employing fault-tolerant building blocks inspired by the edge connection pattern of five varieties of Platonic solids that turns the proposed architecture to a flexible scheme. It also escapes the adoption of any leveling policy for the network equipment, which leads to the efficient usage of all the employed resources. Moreover, to ameliorate the routing efficiency and the network fault-tolerance, two exclusively designed routing algorithms are presented. Both theoretical analysis and simulation results demonstrate that Platonica is highly fault-tolerant and can provide notable network capacity, good average path length, and low latency to support delay-sensitive and data-intensive applications. It can also be considered as a cost-effective and power-efficient network architecture. Altogether, Platonica achieves a good balance among all the major goals of designing cloud data center networks.

Mobile Edge Computing (MEC) provides different storage and computing capabilities within the access range of mobile devices. This moderates the burden of offloading compute/storage- intensive processes of the mobile devices to the centralized cloud data centers. As a result, the network latency is reduced and the quality of service provided for the mobile end users is improved. Different applications benefit from the large-scale deployments of MEC servers. However, the considerable complexity of managing large scale deployments of the sheer number of applications for the millions of mobile devices is a challenge. Recently, Software Defined Networking (SDN) is leveraged to resolve the problem by providing unified and programmable interfaces for managing network devices. Most of the current SDN packet processing services are tightly dependent on the packet classification service. This primary service classifies any incoming packet based on matching a set of specific fields of its header against a flow table. Acceleration of this basic process considerably increases the performance of the SDN-based MEC. In this paper, the hierarchical tree algorithm, which is a packet classification method, is parallelized using popular platforms on a cluster of Graphics Processing Units (GPUs), a cluster of Central Processing Units (CPUs), and a hybrid cluster. The best scenario for the parallel implementation of this algorithm on the CPU cluster is that which combines OpenMP and MPI. In this case, the throughput of the classifier is 4.2 million packets per second (MPPS). On the GPU cluster, two different scenarios have been used. In the first scenario, the global memory is used to store the rules and the Hierarchical-trie of the classifier while in the second scenario we break the filter set in a way that the resulting Hierarchical-trie of each subset could be stored in the shared memory of GPU. According to the results, although the first GPU cluster scenario achieves a throughput of 29.19 MPPS and a speedup 58 times as great as the serial mode, the second scenario is 12 times faster due to using the shared memory. The best performance, however, belongs to the hybrid cluster mode. The hybrid cluster achieves a throughput of 30.59 which is 1.4 MPPS more than the GPU cluster.

The publish-subscribe programming model has been an alternative to the design of data-intensive distributed applications in many domains. Recently, this model has been ported to the domain of Multiprocessor Systems-on-Chips, in which applications must use the underlying Network-on-Chip communication infrastructure effectively due to restrictions on the architecture such as low power consumption and limited memory size. In such a scenario, the publish-subscribe model fulfills some of these requirements while providing high-level access to the network hardware to programmers, thus contributing to software quality. However, the publish-subscribe model relies on a single process dedicated to orchestrating the communication at the application level, the broker. Should a broker process crash, the communication between associated nodes may experience delays, downtime, or even inconsistent data. In extreme cases, communication is definitively ruined. Thus, a recovery strategy for brokers in the publish-subscribe model becomes crucial when the application has safety requirements. In this work, we extend a publish-subscribe protocol to add redundancy to brokers’ sensitive data. Besides, we provide a recovery protocol to recover brokers in case of a failure. We also provide analytical models to estimate the communication overhead of our approach. We validate our approach in two distinct MPSoC platforms. The results show that our approach inserts a small memory footprint to the system while providing minimal system downtime during recovery.

Guaranteeing that large scale networks are always operated as policy goals dictate is critical for network availability, security, and performance. The problem is challenging, as it should not only assure the correctness of policy configurations across various network devices, but also guarantee the faithfulness of policy enforcement on actual packet forwarding behaviors. In this article, we provide a systematic overview of the direction of network verification, which covers both verification of network configurations and assurance of device actual behaviors. In particular, we summarize state-of-the-art designs with focuses from early stateless data planes with switches/routers to more recent stateful ones containing middleboxes. After analyzing and comparing these works, we also specify future directions in the verification of stateful networks.

Debugging a traditional network is notoriously difficult due to network devices’ heterogeneity and protocols’ decentralized nature, but Software-Defined Networking (SDN) is changing this predicament. Recent works have provided very nice approaches for an administrator to perform several fundamental network tests in a single-domain SDN network. However, how to perform these tests securely in multi-domain networks still remains open. In this paper, we study the highly challenging problem of inter-domain forwarding loop test in a SDN environment. We present two novel testing protocols that can be used for inter-domain loop tests. Both protocols are secure in the sense that they protect each domain's private information about its topology and configuration. The first protocol, based on random sampling, is highly efficient with a small error probability diminishing exponentially in the sample size. The second protocol, based on secure set intersection test, guarantees 100 percent accuracy of the result, although not as efficient as the first one. We provide rigorous proofs for the security and accuracy guarantees, and show our protocols have very good efficiency by testing them with real-world network data.

Distributed Denial-of-Service (DDoS) attacks disrupt servers, services and the network, overloading the target resources and denying normal traffic. In order to defend from this attack, mitigation actions usually overprovision and sinkhole malicious traffic. Sooner the attack is detected, better is the mitigation. Hence, we advocate for using a prediction technique aiming to anticipate actions against the possible attack, before it effectively starts. Then, this article contributes to advance the state-of-the-art presenting a distributed architecture that identifies early signals of a possible DDoS attack and detects bots composing a botnet. The architecture identifies the malicious actors (bots) participating in the attack. The bot detection technique is triggered by the prediction of DDoS supported by early signals. Prediction identifies signals of attack on the network before it reaches advanced stages. Based on the metastability theory, it provides unsupervised statistical learning and identifies the imminence of DDoS attacks. The botnet detection is challenging because of the high dimension of data involved and because of resource constraints (memory and processing) in network devices. Network devices are clustered based on features extracted from the traffic and based on the causality between devices. Detection is performed per cluster. Performance evaluations took as input the CTU-13 Czech Republic University, CAIDA and Botnet 2014 datasets, efficiently detecting the bots in the dataset with an accuracy of 99.9%.

The Internet of Things (IoT) has emerged as a promising paradigm to enhance the living standard of human life by employing varied smart devices including smart phones, smart watches, sensors, on-board units and other networking equipment. However, these devices consume a considerable amount of energy to perform their operations that has a significant impact on the environment, product cost and life of the device. Given this fact, energy-efficient solutions for smart environments have gained great attention from researchers and the industrial community. In this context, a novel fog-based multi-level energy-efficient framework for IoT-enabled smart environments has been proposed. To achieve this, the proposed framework adds additional two layers in the existing IoT-fog-cloud architecture - sensors-based energy-efficient hardware layer and policy layer, to monitor the energy consumption and to enable the energy-aware decision making. Initially, the main sources of energy consumption in an IoT-enabled smart environment are identified. Further, the energy requirements of a device to perform a specific task are estimated. Moreover, the alternative devices to perform the same task using less energy are searched out. Finally, a device or a set of devices, to process the job consuming lower energy while ensuring the job requirements, is selected. To validate the proposed framework, four case studies are considered - smart parking, smart hospital specifically ICU, smart agriculture and smart airport. Simulations are conducted using iFogsim toolkit and results show that a significant amount of energy can be conserved by employing the proposed framework.

The incoming 5G technology is expected to proliferate tremendous internet-of-thing (IoT) services with real-time and mobility requirements, which are quite different from the legacy cloud services. Due to the centralized management relying on distant datacenters, cloud computing is short of satisfying the stringent IoT requirements, such as ultra-low latency, mobility, etc. Instead, distributed edge computing, such as fog computing has been coined as a promising approach and has received enormous attention in recent years. In this paper, to optimally provision the huge volume of IoT services with significant diversity, we propose to efficiently organize the leisure network devices in the network edge to form fog networks (fogs), which are then integrated with the cloud to provide storage and computing resources. Specifically, we propose two Integer Linear Programming (ILP) models to solve the fog planning issue under the integrated Cloud-Fog (iCloudFog) framework. In the first ILP model, the objective is to minimize the CAPEX cost caused by planning fogs and the OPEX cost caused by utilizing the planned fogs. In the second ILP model, the objective is to minimize the power consumption while maximizing the number of successfully provisioned IoT tasks on the planned fogs. The proposed ILP models are numerically evaluated by considering different IoT task requirements, such as real-time and mobility. The numerical results show that efficiently planned fogs can help to reduce the planning overhead while satisfying diverse IoT task requirements.

The Internet of things (IoT) is becoming more and more flexible and economical with the advancement in information and communication technologies. However, IoT networks will be ultra-dense with the explosive growth of IoT devices. Network function virtualization (NFV) emerges to provide flexible network frameworks and efficient resource management for the performance of IoT networks. In NFV-enabled IoT infrastructure, service function chain (SFC) is an ordered combination of virtual network functions (VNFs) that are related to each other based on the logic of IoT applications. However, the embedding process of SFC to IoT networks is becoming a big challenge due to the dynamic nature of IoT networks and the abundance of IoT terminals. In this paper, we decompose the complex VNFs into smaller virtual network function components (VNFCs) to make more effective decisions since VNF nodes and IoT network devices are usually heterogeneous. In addition, a deep reinforcement learning (DRL) based scheme with experience replay and target network is proposed as a solution that can efficiently handle complex and dynamic SFC embedding scenarios in IoT. Our simulations consider different types of IoT network topologies. The simulation results present the efficiency of the proposed dynamic SFC embedding scheme.

Raising density and ever-increasing traffic demand within future 5G Heterogeneous Networks (HetNets) will result in huge deployment, expansion and operating costs for upcoming Mobile Backhaul (MBH) networks. Multitenancy and network slicing based on virtualized resources are promising solutions to satisfy MBH network greediness while reducing related expenditures. Nevertheless, there is no appropriate model that fairly distributes costs over multiple virtual operators, and also optimizes physical resource planning. In this article, we introduce a new model of 5G multitenant MBH costs (CapEx and OpEx). Then, we drive a novel pay-as-you-grow and optimization model called Virtual-Backhaul-as-a-Service (VBaaS) as a planning tool optimizing the Project Profit Margin (PPM) while considering the Total-Cost-of-Ownership (TCO) and the yearly generated Return-on-Investment (ROI). We also formulate an MNO pricing game (MPG) for TCO optimization to calculate the optimal Pareto-Equilibrium pricing strategy for offered Tenant Service Instances (TSI). Finally, we compare CapEx, OpEx, TCO, ROI and PPM for a specific use-case known in the industry as CORD project using Traditional MBH (T-MBH) versus Virtualized MBH (V-MBH) as well as using randomized versus Pareto-Equilibrium pricing strategies. Numerical results show more than three times increase in network profitability using our proposed solutions compared with Traditional MBH (T-MBH).

By means of network function virtualization (NFV), dedicated proprietary network devices can be implemented as software and instantiated flexibly on common-off-the-shelf servers, in the form of virtual network functions (VNF). NFV can bring great cost reduction as well as operation flexibility. However, it also brings new problems, one of which is how to meet the availability of network services in the VNF deployment process, because of the error prone nature of software. The availability aware VNF deployment problem has attracted attention by academics, and reserving redundancy has been treated as the de facto technology. Compared with traditional backup schemes for physical machines, resource orchestration in NFV is more flexible and the characteristics of software should be considered to improve resource utilization efficiency. Based on the above considerations, in this paper we further study the availability aware VNF deployment problem in datacenter networks. To improve the resource utilization efficiency, the sharing mechanism of redundancy and multi-tenancy technology are taken into account. Then we formulate the problem mathematically and propose a joint deployment and backup scheme (JDBS). Finally, we conduct a numerical simulation in detail and compare it with four contrasting schemes in the existing literature. The simulation results show that JDBS is obviously superior to the contrasting schemes and can save about 40% resources at most.

Theallreduceoperation is one of the most commonly used communication routines in distributed applications. To improve its bandwidth and to reduce network traffic, this operation can be accelerated by offloading it to network switches, that aggregate the data received from the hosts, and send them back the aggregated result. However, existing solutions provide limited customization opportunities and might provide suboptimal performance when dealing with custom operators and data types, with sparse data, or when reproducibility of the aggregation is a concern. To deal with these problems, in this work we design a flexible programmable switch by using as a building block PsPIN, a RISC-V architecture implementing the sPIN programming model. We then design, model, and analyze different algorithms for executing the aggregation on this architecture, showing performance improvements compared to state-of-the-art approaches.

Finding a more efficient use of energy is an important problem that needs attention. Compared with the traditional power grid, a smart grid can monitor users’ electricity situation and electricity consumption instantly. However, it involves many problems of deploying network equipment. Consequently, it is vital to promote smart grids by collecting data from smart meters efficiently and keeping costs low. In this article, we propose a two-stage method of data collection for smart grids. The main contribution of this paper is to lower the number of data aggregation points (DAPs) so that the cost can be reduced. By using the K-means method, an entire smart grid can be divided into many smaller parts. In addition, the needs of transmitting and receiving data in the entire smart grid can be met by installing the least number of DAPs. Finally, the simulations show that the proposed two-stage method of data collection can use fewer DAPs to collect data than other methods which use one-stage methods, so the proposed scheme is more cost-effective.

In-network computing via smart networking devices is a recent trend for modern datacenter networks. State-of-the-art switches with near line rate computing and aggregation capabilities are developed to enable, e.g., acceleration and better utilization for modern applications like big data analytics, and large scale distributed and federated machine learning. We formulate and study the problem of activating a limited number of in-network computing devices within a network, aiming at reducing the overall network utilization for a given workload. Such limitations on the number of in-network computing elements per workload arise, e.g., in incremental upgrades of network infrastructure, and are also due to requiring specialized middleboxes, or FPGAs, that should support heterogeneous workloads, and multiple tenants. We present an optimal and efficient algorithm for placing such devices in tree networks with arbitrary link rates, and further evaluate our proposed solution in various scenarios and for various tasks. Our results show that having merely a small fraction of network devices support in-network aggregation can lead to a significant reduction in network utilization. Furthermore, we show that various intuitive strategies for performing such placements exhibit significantly inferior performance compared to our solution, for varying workloads, tasks, and link rates.

In recent years, Internet Service Providers (ISPs) are expected to enable various practical services. To meet the requirements of services, the infrastructure of networks has become more and more complex. In telecom networks, the complex infrastructure implies that it would be difficult to analyze the root causes and to locate the faults. In the telecom companies, network maintenance staffs need to spend a lot of time to trace the root cause and solve the network problems. An intelligent fault location approach allows ISPs to be cost-effective, and can assist humans in decision-making and increase automation. To automatically locate the faults, we apply both the Ant Colony Optimization (ACO) algorithm and fuzzy logic methods, and the main contributions of this paper are threefold: (1) we apply the pheromones of the ACO algorithm to quantify the risks that network devices might fail; (2) based on the risks, we leverage fuzzy logic, including the fuzzy relation matrix and the max-min composition method, to infer the fault location; (3) for improving autonomous network, we implemented and evaluated the proposed intelligent fault location approach using the real data in telecom networks.

Internet of Things (IoT) has been developed as a heterogeneous environment that contains network devices with limited resources. The application of IoT principles in the smart city domain creates new opportunities and requires diligent implementation mechanisms for optimal resource utilization. With time, the IoT applications tend to generate and forward a huge amount of data in the smart cities and require a real-time response from the servers. Due to this, the traditional cloud computing architecture is unable to handle the latency-sensitive applications efficiently, and hence, the FoG architecture has been widely implemented with IoT devices to efficiently retrieve or forward the data. For the comprehensive utilization of the resources in the FoG systems-based smart cities, various energy-aware resource allocation schemes have been discussed in this chapter. The schemes suggest different mechanisms to access the required contents with minimal energy consumptions for the applications that are used in smart cities.

Data classification within the network brings significant benefits in reaction time, servers offload and power efficiency. Still, only very simple models were mapped to the network. In-network classification will not be useful unless we manage to map complex machine learning models to network devices. We present Planter, an algorithm that maps a variety of ensemble models, such as XGBoost and Random Forest, to programmable switches. By overlapping trees within coded tables, Planter manages to map ensemble models to switches with high accuracy and low resource overhead.

Over the years, the need for communication networks capable of providing an ever-increasing set of services has grown. In order to satisfy user requirements and provide guarantees of reliability of the network itself, efficient techniques are required for analysis, evaluation and design. For this reason, the need arises to have models able to represent the peculiar characteristics of network traffic and to produce reliable predictions of its behavior in an adequate period of time. Therefore, network traffic prediction plays an important role by supporting many practical applications, ranging from network planning and provisioning to security. Several works so far have focused on building app-specific models. However, this choice produces multiple models that need to be properly managed and deployed across network devices. Therefore, in this paper, we explore different training strategies to reduce the number of models, adopting the Markov Chains to model mobile video apps traffic at packet-level. We discuss and experimentally evaluate the prediction effectiveness of the proposed approaches by comparing the performance of app models with models trained on a specific category of video apps and a model trained on the mix of all video traffic.

MPI provides nonblocking point-to-point and one-sided communication models to help applications achieve communication and computation overlap. These models provide the opportunity for MPI to offload data transfer to low level network hardware while the user process is computing. In practice, however, MPI implementations have to often handle complex data transfer in software due to limited capability of network hardware. Therefore, additional asynchronous progress is necessary to ensure prompt progress of these software-handled communication. Traditional mechanisms either spawn an additional background thread on each MPI process or launch a fixed number of helper processes on each node. Both mechanisms may degrade performance in user computation due to statically occupied CPU resources. The user has to fine-tune the progress resource deployment to gain overall performance. For complex multiphase applications, unfortunately, severe performance degradation may occur due to dynamically changing communication characteristics and thus changed progress requirement. This paper proposes a novel Dynamic Asynchronous Progress Stealing model, called Daps, to completely address the asynchronous progress complication. Daps is implemented inside the MPI runtime. It dynamically leverages idle MPI processes to steal communication progress tasks from other busy computing processes located on the same node. The basic concept of Daps is straightforward; however, various implementation challenges have to be resolved due to the unique requirements of interprocess data and code sharing. We present our design that ensures high performance while maintaining strict program correctness. We compare Daps with state-of-the-art asynchronous progress approaches by utilizing both microbenchmarks and HPC proxy applications.

Network Softwarization, the process by which network equipment's are being replaced by software running in a cloud environment, is playing an important role in the transformation of next-generation networks starting with 5G. In this environment, the use of AI/ML aided network automation to perform management tasks is vital to provide a more reliable and cost-effective network. An aspect in which AI/ML can play an important role is on the quick diagnose and track of root causes of anomalies. In this paper, we propose a system capable of automating root cause analysis in a cellular network scenario. By collecting log files from negative-impacting events occurring throughout the whole network levels, we are able to find the correlation between them according to a series of defined rules and track down the root cause using a graph-based approach to dependencies

The world is entering into the era of Big Data where computer networks are an essential part. However, the current network architecture is not very convenient to configure such leap. Software defined network (SDN) is a new network architecture which argues the separation of control and data planes of the network devices by centralizing the former in high level, centralised devices and efficient supervisors, called controllers. This paper proposes a mathematical model that helps optimizing the locations of the controllers within the network while minimizing the overall cost under realistic constrains. Our method includes finding the minimum cost of placing the controllers; these costs are the network latency, controller processing power and link bandwidth. Different types of network topologies have been adopted to consider the data profile of the controllers, links of controllers and locations of switches. The results showed that as the size of input data increased, the time to find the optimal solution also increased in a non-polynomial time. In addition, the cost of solution is increased linearly with the input size. Furthermore, when increasing allocating possible locations of the controllers, for the same number of switches, the cost was found to be less.

With the emergence of programmable network devices that match the performance of fixed function devices, several recent projects have explored in-network computing, where the processing that is traditionally done outside the network is offloaded to the network devices. In-network computing has typically been applied to network functions (e.g., load balancing, NAT, and DNS), caching, data reduction/aggregation, and coordination/consensus functions. In some cases it has been used to accelerate stream-processing tasks that involve small payloads and simple operations. In this work we focus on leveraging in-network computing for stream processing of scientific datasets with large payloads that require complex operations such as floating-point computations and logarithmic functions. We demonstrate in-network computing for a real-world scientific application performing streaming normalization of a 2-D image from a light source experiment. We discuss the challenges we encountered and potential approaches to address them.

Timeliness and reliability are two major requirements of industrial applications. Hence, these two requirements should be carefully taken into account in design of a smart factory. Using a hybrid Cloud (combining a public Cloud with a private Cloud) is a well-known method to enable Cloud computing to provide reliability and timeliness. However, for strict time-sensitive applications using a hybrid Cloud is not enough to guarantee meeting their hard-deadlines. Supplying an intermediate computing layer –called Fog– between the factory and the Cloud is a promising solution to deal with reliability and latency requirements of strict time-sensitive applications. Integration of a Cloud data center, a private local Cloud, Fog nodes, and edge nodes constitute a very complex multilayer computing model in a smart factory. The goal of this paper is to provide a resource provisioning scheme for partitioning of a given workload among these multiple computing layers subject to reliability and real-time requirements. Partitioning of the workload can provide us prominent design decisions specifying how much computing resources are required to develop a local private Cloud in cooperating with Fog nodes like networking devices, how large should the minimum communication bandwidth be between the Fog and the public Cloud data center, and how many replicas for each application are required to satisfy the reliability requirement of the considered application. To evaluate the proposed method, we have conducted a set of experiments ranging from small scale to large scale scenarios. The results indicate that in 86 percent of the small scale experiments, the proposed algorithm exactly found the optimal solution achieved by a Branch and Bound (BB) based exhaustive search algorithm, while there is an improvement of around 85 percent on the execution time of our proposed method compared to the exhaustive search method.

Inter-core crosstalk causes serious signal impairment in a multi-core fiber (MCF) optical network. With the extensive usage of data centers, traffic demands become increasingly asymmetric, leading to significant overall capacity wastage if the network is designed with bidirectional links. To alleviate these effects in an MCF optical network, we consider assigning the fiber cores asymmetrically, in a counter-propagating manner. This not only reduces the inter-core crosstalk between counter-propagating fiber cores significantly but also allows the flexibility of assigning different numbers of fiber cores in the opposite directions of a fiber link, thereby avoiding network capacity wastage arising from bidirectional traffic demand asymmetry. To evaluate the benefits of the proposed strategy, we consider the routing, spectrum, and core assignment (RSCA) problem for the MCF optical network. An integer linear programming (ILP) model and an auxiliary graph based heuristic algorithm are developed to optimize network spectrum resource utilization. Simulation studies show the effectiveness of the proposed crosstalk-aware core counter-propagation strategy, which can significantly outperform its counterpart, i.e., the co-propagation scheme, in terms of the total number of MCFs used and average inter-core crosstalk. In addition, the proposed RSCA heuristic algorithm is efficient to perform close to the ILP model, which can minimize the number of MCFs used and crosstalk between neighboring cores.

Software Defined Network (SDN) has become one of the most preferred solutions for the management of large-scale complex networks. The network policies in the large-scale network are difficult to embed on entire network devices simultaneously, whereas in SDN these policies can be embedded on the top of the network. The SDN network is divided into two parts which are vertically integrated to form the entire network. The policies and control mechanism are handled by the top layer called the control plane, whereas the data forwarding is performed at the bottom layer called the data plane. Thus the SDN forms a logically centralized network. In this paper, we present a comprehensive survey on the various issues, challenges, and solutions in the designing, implementation, performance, and verification of SDN. The network design emphasized on scalability, fault-tolerance, flexibility, and elasticity. Further implementation issues discuss resource management and virtualization. Similarly, latency, efficiency, and consistency discuss under-performance measure. Lastly, we focus on security and debugging in verification of SDN. We elaborate all the above issues and further discuss the different solutions that exist and discussed the future research trends in SDN.

With the development of data center around the world, its energy consumption becomes an increasingly urgent issue, and green data center (GDC) emerges. Most modern data centers are operated in terms of peak load, with large scales. However, many servers and network devices are idle or light-loaded for a long time, which causes energy waste. Therefore, it is necessary to carry out energy saving in GDC. The work studied the optimization of energy consumption of switches, proposing an energy-saving strategy based on artificial bee colony (ABC) for Fat Tree. The simulation results showed the proposed algorithm has obvious effects on energy saving and completion time in large and small traffic, which embodies the superiority and effectiveness of the algorithm.

Software-defined networking (SDN) has evolved as an effective platform for future Internet due to its capability of configuring the network dynamically with varying requirements. It has been observed that the load and energy requirement of SDN devices increase significantly with the growth of communication networks. Therefore, there is a need for efficient modeling of SDN controller that can balance the load as well as optimize the energy consumption by the devices. In this paper, we present an energy-efficient load distribution framework; controller system model for efficient load distribution and routing of traffic that objectively optimize the energy consumption in the network. Our model balances load according to the heterogeneous traffic demands as well as reduces energy consumption by introducing energy-efficient routing algorithm selection procedure. The load balancing scheme is drifted by switch migration technique for multiple controllers simultaneously, whereas the novelty of energy-efficient routing lies on sleep and active mode of network devices. We present interaction between load-balancing scheme and energy-efficient routing towards the network’s performance enhancement. The efficacy of our proposed controller system model is justified with extensive simulation results that show approximately 25% reduction of energy consumption and approximately 20% performance increment. Our proposed model is applicable to real-life network environment satisfying the standards of green communication.

This paper introduces a novel approach for detecting the participation of a protected network device in flooding-based Distributed Denial of Service attacks. With this purpose, the traffic flows are inspected at source-side looking for discordant behaviors. In contrast to most previous solutions, the proposal assumes the non-stationarity and heterogeneity inherent in the emergent communication environment. In particular, the approach takes advantage of the monitorization and knowledge acquisition capabilities implemented in the SELFNET (H2020-ICT-2014-2/671672) project, which facilitates its implementation as a self-organizing solution on 5G mobile networks. Monitorization, feature extraction and knowledge acquisition tasks are carried out on centralized control plane, hence the proposed architecture minimizes the impact on operational performance and prompts the end-points mobility. The preliminary results observed when considering different metrics, adjustment parameters, and a dataset with traffic observed in 61 real devices proven efficiency when distinguishing normal activities from DDoS behaviors of different intensity. With an optimal granularity selection, the highest AUC reached values close to 1.0 when measured under the most intense attacks, hence demonstrating optimal TPR and FPR relationships by adapting to the instantiated use cases.

As the volume of data keeps growing rapidly, more and more storage devices, servers and network devices are continuously added into data centers to store, manage and analyze the data. The industry experience indicates that, instead of a huge number of servers added at a time, the data center network also expands gradually by adding a small number of servers from time to time. As a result, how to achieve an incremental scalability is becoming a very important challenge in designing modern data center network architectures in order to maintain the topological properties unchanged when the size of data centers grows. In this paper, we propose a new type of data center network architecture named HSDC (High Scalability Data Center Network Architecture) based on the hypercube network. The HSDC is constructed by using $m$m-port switches and 2-port servers. The fault-tolerant routing algorithm designed in this paper for HSDC can be executed on any vertex and is able to construct a path between any pair of vertices. In order to achieve an incremental scalability, we further propose three types of incomplete HSDC structures that allow gradually adding servers into the structures, while maintaining all the topological properties. The simulation experiments and performance results demonstrate that the throughput of HSDC is comparable to that of Fat-Tree, BCube and DCell. Furthermore, the analysis results indicate that HSDC strikes a good balance among diameter, bisection width, incremental scalability, cost and energy consumption in contrast to the state-of-the-art data center network architectures.

Network function virtualization (NFV) has been considered as a promising technology for future Internet to increase the network flexibility, accelerate the service innovation, and reduce the Capital Expenditures and Operational Expenditures costs through migrating network functions from dedicated network devices to commodity hardware. Recent studies reveal that although this migration of network function brings the network operation unprecedented flexibility and controllability, NFV-based architecture suffers from serious performance degradation compared with traditional service provisioning on dedicated devices. In order to achieve a comprehensive understanding of the service provisioning capability of NFV, this paper proposes a novel analytical model based on Stochastic Network Calculus (SNC) to quantitatively investigate the end-to-end performance bound of the NFV networks. To capture the dynamic and on-demand NFV features, both the non-bursty traffic, e.g., the Poisson process, and the bursty traffic, e.g., the Markov Modulated Poisson Process, are jointly considered in the developed model to characterize the arriving traffic. To address the challenges of resource competition and end-to-end NFV chaining, the property of convolution associativity and leftover service technologies of SNC are exploited to calculate the available resources of the Virtual Network Function nodes in the presence of multiple competing traffic and transfer the complex NFV chain into an equivalent system for performance derivation and analysis. Both the numerical analysis and extensive simulation experiments are conducted to validate the accuracy of the proposed analytical model. Results demonstrate that the analytical performance metrics match well with those obtained from the simulation experiments and numerical analysis. In addition, the developed model is used as a practical and cost-effective tool to investigate the strategies of the service chain design and resource allocations in the NFV networks.