Efficient Network Resource Management Research Articles

Neural Network Multipath Routing in Software Defined Networks Based on Genetic Algorithm

Currently, a wide demand for the implementation and use of various cloud solutions is a modern trend and the driving force behind the development of network technologies. The growth of cloud application services delivered through data centers with varying network traffic needs demonstrates the limitations of traditional routing and load balancing methods. The combination of the advantages of software defined networks (SDN) technology and artificial intelligence (AI) methods ensures efficient management and operation of computer network resources. The paper proposes an approach to neural network multipath routing in SDN based on a genetic algorithm. The architecture and model of an artificial neural network has been developed to solve the problem of multipath routing in the SDN, which is able to predict the shortest paths based on the metrics of communication links. To optimize the hyperparameters of the neural network model, it is proposed to use a modified genetic algorithm. A visual software system SDNLoadBalancer has been developed and an experimental SDN topology has been designed, which makes it possible to study in detail the processes of neural network multipath routing in SDN based on the proposed approach. The obtained results show that the proposed neural network model has the ability to predict routes with high accuracy in real time, which makes it possible to implement various load balancing schemes in order to increase performance of SDN.

Quality of service management method in а heterogeneous wireless network using Big Data technology аnd mobile QoE application

Today, with the increasing number of business users and their changing Quality of Service (QoS) requirements, there is a problem for mobile network operators to provide individual acceptable Quality of Experience (QoE) to their customers due to the lack of flexibility, agility, and scalability of traditional networks. For this reason, the latest scientific works are increasingly focusing on resource management, considering users' QoE, replacing the traditional system-oriented approach based on QoS. However, this problem has not been fully solved so far, as there are no approaches that take into account the user's intentions regarding the individualization of the specific types of services at a particular time for heterogeneous mobile networks. Thus, the development of means of feedback between the user and the operator on the adaptation of the QoE level or their intentions under changing user requirements is one of the important and unresolved tasks for future networks at the moment. A conceptual model for the construction of an Intent-Based Software-Defined Heterogeneous Network (IBSDHN) is proposed. This model ensures efficient management of shared network resources, adapting to business users' changing QoS requirements. The prototype of a mobile and operator application for adaptive client-oriented service delivery in a heterogeneous wireless network is developed, which allows obtaining the ordered QoE based on the feedback between the user and the operator. A method for QoS management in a heterogeneous wireless network using Big Data technology and a unique mobile (QoE) application is proposed. In contrast to the known, this method analyzes ordered QoE scores of service from users and searches for the optimal radio access network that meets these requirements based on the predicted QoE values. The experimental study proved that using the proposed method in a heterogeneous wireless network would reduce the number of dissatisfied users with the QoS up to 60%.

HFedDI: A novel privacy preserving horizontal federated learning based scheme for IoT device identification

As the number of IoT devices that are getting connected to the Internet is increasing, there is a need to automatically detect the IoT devices connected to the network for efficient network resource management and planning, identification of security attacks and anomalies in the network. Centralized machine learning techniques for device identification lead to the use of excessive communication bandwidth and can at times lead to privacy issues as data has to be transported to the central location. In this paper, we propose HFedDI- a novel horizontal learning based federated learning scheme for IoT device identification. The proposed federated learning technique is scalable as an edge device that does device identification can identify the devices which were never connected to it previously due to the achieved generalization accuracy as a result of model updates from other edge devices. Our work has indicated significant performance improvement in device identification across IID scenarios, and non-IID (specifically tested for data and label distribution skew) scenarios when tested on three publically available datasets. The proposed device identification technique is privacy preserving and is promising as the existing work in literature which utilized federated learning for other applications has indicated very poor results under non-IID label skew scenarios.

Low Rate DDoS Detection Using Weighted Federated Learning in SDN Control Plane in IoT Network

The Internet of things (IoT) has opened new dimensions of novel services and computing power for modern living standards by introducing innovative and smart solutions. Due to the extensive usage of these services, IoT has spanned numerous devices and communication entities, which makes the management of the network a complex challenge. Hence it is urgently needed to redefine the management of the IoT network. Software-defined networking (SDN) intrinsic programmability and centralization features simplify network management, facilitate network abstraction, ease network evolution, has the potential to manage the IoT network. SDN’s centralized control plane promotes efficient network resource management by separating the control and data plane and providing a global picture of the underlying network topology. Apart from the inherent benefits, the centralized SDN architecture also brings serious security threats such as spoofing, sniffing, brute force, API exploitation, and denial of service, and requires significant attention to guarantee a secured network. Among these security threats, Distributed Denial of Service (DDoS) and its variant Low-Rate DDoS (LR-DDoS), is one of the most challenging as the fraudulent user generates malicious traffic at a low rate which is extremely difficult to detect and defend. Machine Learning (ML), especially Federated Learning (FL), has shown remarkable success in detecting and defending against such attacks. In this paper, we adopted Weighted Federated Learning (WFL) to detect Low-Rate DDoS (LR-DDoS) attacks. The extensive MATLAB experimentation and evaluation revealed that the proposed work ignites the LR-DDoS detection accuracy compared with the individual Neural Networks (ANN) training algorithms, existing packet analysis-based, and machine learning approaches.

Dynamic Radio Access Selection and Slice Allocation for Differentiated Traffic Management on Future Mobile Networks

The development of future wireless networks focuses on providing services with strict, dynamic, and diverse quality of service (QoS) requirements. In this sense, the network slicing paradigm arises as a critical piece on the efficient allocation and management of network resources, allowing for dividing the network into several logical networks with specific functionalities and performance. This paper aims at finding the best combination of access network and network slices over a heterogeneous environment to fulfill users’ requests and optimize network resources usage. We propose the Dynamic radio Access selection and Slice Allocation (DASA) algorithm, flexibly adapted to network conditions, user priorities, and mobility behavior. DASA is based on a multi-attribute decision making (MADM) and analytical hierarchy process (AHP) to face the complex problem of network selection. Moreover, it uses a cooperative game theory approach to handle load balancing during overload situations. This work presents an integral solution that combines software-defined network (SDN) and network function virtualization (NFV) technologies to improve network performance and user satisfaction. DASA algorithm is evaluated through network-level simulations, focusing on flexibility and the effective utilization of network resources during network selection and load balancing mechanisms.

A fuzzy-based fast routing algorithm with guaranteed latency-throughput over software defined networks

With the development of online communications over the Internet, the network is required to guarantee the Quality of Service (QoS) requirements for different traffic flows. High quality and fast communication can be provided by Software-Defined Networks (SDN). SDN is a promising architecture for efficient management of network resources by separating control logic from data planes. In order to ensure the QoS requirements of traffic flows in SDN, this paper proposes a Fuzzy-based fast and efficient Routing algorithm with guaranteed Latency-Throughput (FRLR). The FRLR develops the dynamic routing problem in SDN using a traffic flow classification mechanism based on a fuzzy system. The purpose of the fuzzy system is to efficiently distribute resource allocation to improve the handling of future requests. In addition, the FRLR classifies traffic flows based on latency to reduce computational complexity. Besides, FRLR can effectively direct traffic flows and reduce network energy consumption by identifying critical links and selecting appropriate paths. Simulations on standard topologies with different scenarios confirm the better performance of FRLR by guaranteeing QoS requirements. The FRLR classifies traffic flows efficiently and offers significant improvements in the scalability and throughput of the network. Also, FRLR has provided superior results compared to some state-of-the-art algorithms such as MDA, MDWCRA, FARP, ECMP, and QI-RM. Experimental results show that FRLR works on average between 7% and 24% better than other algorithms.

Confidence Aware Deep Learning Driven Wireless Resource Allocation in Shared Spectrum Bands

Deep learning (DL) driven proactive resource allocation (RA) is a promising approach for the efficient management of network resources. However, DL models typically have a limitation that they do not capture the uncertainty due to the arrival of new unseen samples with a distribution different than the data distribution available at DL model-training time, leading to wrong resource usage predictions. To address this, we propose a confidence aware DL solution for the robust and reliable predictions of wireless channel utilization (CU) in shared spectrum bands. We utilize an encoder-decoder based Bayesian DL model to generate prediction intervals which capture the uncertainties in wireless CU. We use the CU predictions to design a novel metric score which in turn is utilized to make an adaptive RA algorithm. We show that a DL model capturing uncertainty in CU can achieve higher data rates for a wireless network. Both DL driven predictions and RA models are tested using synthetic data as well as real CU data collected in the University of Oulu. Using analytical and simulations results, we also study the stability of the proposed RA algorithm and show that it converges to a Nash equilibrium (NE). Our results reveal that the proposed algorithm converges to an NE under <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2N $ </tex-math></inline-formula> iterations where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$N $ </tex-math></inline-formula> is the number of network access points.

A lightweight model with spatial–temporal correlation for cellular traffic prediction in Internet of Things

Accurate cellular traffic prediction becomes more and more critical for efficient network resource management in the Internet of Things (IoT). However, high-accuracy prediction results are usually accompanied by high computational capacity requirements. Although many lightweight neural network models have been proposed, some lightweight mechanisms will easily destroy the features of the raw data. Not all lightweight mechanisms are suitable for network traffic prediction. Therefore, this study proposes and optimizes an input data conversion method to extract the features of spatio-temporal dependencies based on convolutional neural network (CNN) architecture. In addition, we also propose a lightweight neural network model to reduce the computational cost for cellular traffic prediction problem and use mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) to evaluate the prediction accuracy. The experimental results show that the proposed model is better than CNN, ConvLstm, and Densenet as well as can greatly reduce the parameters of the neural network while maintaining prediction accuracy.

Learning to Schedule Network Resources Throughput and Delay Optimally Using Q+-Learning

As network architecture becomes complex and the user requirement gets diverse, the role of efficient network resource management becomes more important. However, existing throughput-optimal scheduling algorithms such as the max-weight algorithm suffer from poor delay performance. In this paper, we present reinforcement learning-based network scheduling algorithms for a single-hop downlink scenario which achieve throughput-optimality and converge to minimal delay. To this end, we first formulate the network optimization problem as a Markov decision process (MDP) problem. Then, we introduce a new state-action value function called Q <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> -function and develop a reinforcement learning algorithm called Q <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> -learning with UCB (Upper Confidence Bound) exploration which guarantees small performance loss during a learning process. We also derive an upper bound of the sample complexity in our algorithm, which is more efficient than the best known bound from Q-learning with UCB exploration by a factor of γ <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> where γ is the discount factor of the MDP problem. Finally, via simulation, we verify that our algorithm shows a delay reduction of up to 40.8% compared to the max-weight algorithm over various scenarios. We also show that the Q <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">+</sup> -learning with UCB exploration converges to an ε-optimal policy 10 times faster than Q-learning with UCB.

Resource Allocation for Hybrid Visible Light Communications (VLC)-WiFi Networks

Visible Light Communication (VLC) is becoming a promising technology of wireless communication, due to high data rate, low cost, and immunity to Radio Frequency (RF) interference. However, the VLC systems often suffer from service disruptions because of the limited coverage of light. Therefore, the heterogeneous networks of VLC and WiFi are proposed to achieve high-speed indoor communication and seamless coverage. This paper investigates an efficient heterogeneous network resource management algorithm, which is not only significant to integrate the current network resources but also has practical significance to improve the performance of heterogeneous networks. We establish a distributed joint resource management system model and propose two kinds of resource management allocation strategies to allocate bandwidth among the users with maximum fairness. The simulation results show the feasibility of the heterogeneous network and the effectiveness of the algorithm.

Automation of 5G Network Slice Control Functions with Machine Learning

5G communication networks will be complex due to the emergence of an unprecedented huge number of new types of connected devices and services. Moreover, the on-demand creation of virtual network slices, each suitable for a different application, is posing challenges to the efficient management of network resources, while optimally satisfying the quality of service requirements in time-varying workloads and network conditions. This article, which is tutorial in nature, introduces 5G network slices (from the point of view of the non-wireless part of the network) and elaborates the necessity of automation of network functions related to the design, construction, deployment, operation, control, and management of network slices. It revisits machine learning techniques applicable to the automation of network functions. It then presents a machine-learning-based framework for the operation and control of network slices by continuously monitoring workload, performance, and resource utilization, and dynamically adjusting the resources allocated to network slices. Preliminary results of workload prediction accuracy obtained from the analysis of real-life data collected from a web server are also reported.

Joint Optimization of Caching, Computing, and Radio Resources for Fog-Enabled IoT Using Natural Actor–Critic Deep Reinforcement Learning

The cloud-based Internet of Things (IoT) develops rapidly but suffer from large latency and backhaul bandwidth requirement, the technology of fog computing and caching has emerged as a promising paradigm for IoT to provide proximity services, and thus reduce service latency and save backhaul bandwidth. However, the performance of the fog-enabled IoT depends on the intelligent and efficient management of various network resources, and consequently the synergy of caching, computing, and communications becomes the big challenge. This paper simultaneously tackles the issues of content caching strategy, computation offloading policy, and radio resource allocation, and propose a joint optimization solution for the fog-enabled IoT. Since wireless signals and service requests have stochastic properties, we use the actor–critic reinforcement learning framework to solve the joint decision-making problem with the objective of minimizing the average end-to-end delay. The deep neural network (DNN) is employed as the function approximator to estimate the value functions in the critic part due to the extremely large state and action space in our problem. The actor part uses another DNN to represent a parameterized stochastic policy and improves the policy with the help of the critic. Furthermore, the Natural policy gradient method is used to avoid converging to the local maximum. Using the numerical simulations, we demonstrate the learning capacity of the proposed algorithm and analyze the end-to-end service latency.

Smart Service-Aware Wireless Mixed-Area Networks

Smart services such as smart home, smart grid, smart transportation, and so on, are enabled with advanced communications networks, data analysis and control. However, challenges exist in designing a supporting network with optimal resource management while achieving quality- of-experience for a smart service. In this article, we propose a unified SA-WMN architecture that can be applied to different smart services. The core of the SA-WMN architecture is a smart service-aware network controller. The mega network controller has a service adaptation module, network cognition module and resource provisioning module that can achieve cost-effective wireless-mixed area network deployment, and efficient network resource management of SA-WMN. The proposed SA-WMN architecture is expected to close the gap on wireless network design, improve energy efficiency and enhance spectrum efficiency and access, in a smart service- aware paradigm. It will pave the way for realizing next-generation wireless network based smart services.

Energy-Efficient Routing Mechanism for Mobile Sink in Wireless Sensor Networks Using Particle Swarm Optimization Algorithm

One of the most effective approaches to increase the lifetime of wireless sensor networks (WSNs), is the use of a mobile sink to collect data from sensor. In WSNs, mobile sinks implicitly help achieving uniform energy-consumption and provide load-balancing. In this approach, some certain points in the sensors field should be visited by the mobile sink. The optimal selection of these points which are also called rendezvous points is a NP-hard problem. Since hierarchical algorithms rely only on their local information to select these points, thus the probability of selecting an optimal node as rendezvous point will be very low. To address this problem, in this paper, a new method called particle swarm optimization based selection (PSOBS) is proposed to select the optimal rendezvous points. By applying PSO, the proposed method is capable of finding optimal or near-optimal rendezvous points to efficient management of network resources. In the proposed method, a weight value is also calculated for each sensor node based on the number of data packets that it receives from other sensor nodes. The proposed method was compared with weighted rendezvous planning based selection (WRPBS) algorithm based on some performance metrics such as throughput, energy consumption, number of rendezvous points and hop count. The simulation results show the superiority of PSOBS as compared with WRPBS, but it increases the packet loss rate in comparison with WRPBS.

Column generation algorithms for virtual network embedding in flexi-grid optical networks.

Network virtualization provides means for efficient management of network resources by embedding multiple virtual networks (VNs) to share efficiently the same substrate network. Such virtual network embedding (VNE) gives rise to a challenging problem of how to optimize resource allocation to VNs and to guarantee their performance requirements. In this paper, we provide VNE algorithms for efficient management of flexi-grid optical networks. We provide an exact algorithm aiming to minimize the total embedding cost in terms of spectrum cost and computation cost for a single VN request. Then, to achieve scalability, we also develop a heuristic algorithm for the same problem. We apply these two algorithms for a dynamic traffic scenario where many VN requests arrive one-by-one. We first demonstrate by simulations for the case of a six-node network that the heuristic algorithm obtains very close blocking probabilities to exact algorithm (about 0.2% higher). Then, for a network of realistic size (namely, USnet) we demonstrate that the blocking probability of our new heuristic algorithm is about one magnitude lower than a simpler heuristic algorithm, which was a component of an earlier published algorithm.

Wavelength Allotment for All-to-All Broadcast in WDM Optical Modified Linear Array for Reliable Communication

All-to-all broadcast in a network is to distribute an exclusive message from each node to every other node. It is a vital requirement in high performance computing and communication applications including enterprises operating data centers. It is also a core issue in telecommunication network where each node in the network needs to gather information about all the other nodes for efficient management of network resources. WDM optical networks act as the backbone for telecommunication networks. Wavelengths are precious resources in WDM optical networks and its usage must be limited to reduce network cost and complexity. Optical linear array finds application in LAN, WAN and MAN and its topology needs to be modified to reduce wavelength requirements and hop count. In this paper, a linear array network is modified by additionally joining alternate nodes. For such a modified linear array, the minimum wavelength number needed to support all-to-all broadcast is derived by grouping non overlapping connections on a common wavelength. The results obtained for modified linear array shows that the wavelength number needed is approximately reduced by half when compared with that of a linear array.

Optimized provisioning of SDN-enabled virtual networks in geo-distributed cloud computing datacenters

Cloud computing provides on-demand IT services via large distributed datacenters over high-speed networks. Virtualization, a key cloud computing technology, allows service providers to offer computing services in cloud environments without platform compatibility discrepancies. The recent proliferation of cloud computing has rekindled interest in network virtualization. Thus, network virtualization is emerging as a polymorphic approach for the future Internet that will facilitate the use of shared resources. Virtual network provisioning is considered to be a main resource allocation challenge in any virtualized network environment. Software-defined networking (SDN) imparts flexibility to a network by removing the control layer from the data transfer layer of the network and moving it to the control plane. Network virtualization is further employed to share physical infrastructure to enable multiple service providers to access the network. Flexible access requires efficient management of network resources; the SDN control plane can be used for efficient management of virtual networks. In this study, we formulate virtual network provisioning in SDN-enabled, geographically distributed cloud computing datacenters as a mixed integer linear programming (MILP) problem. The formulation of the proposed optimized virtual network provisioning (OVNP) model is studied by means of simulations. The performance of the proposed approach is measured against enhanced network cloud provisioning (ENCP), our previous research, and other recognized research focused on the ratio of successfully provisioned requests and the efficiency of resource utilization The results verify the effectiveness of the proposed approach.

A Multi-Pronged Approach to Adaptive and Context Aware Content Dissemination in VANETs

Content dissemination in Vehicular Ad-hoc Networks (VANETs) has the potential to enable a myriad of applications, ranging from advertising, traffic and emergency warnings to infotainment. This variety in applications and services calls for mechanisms able to optimize content storing, retrieval and forwarding among vehicles, without jeopardizing network resources. Content Centric Networking (CCN), takes advantage of inherent content redundancy in the network in order to decrease the utilization of network resources, improve response time and content availability, coping efficiently with some of the effects of mobility. Floating Content (FC), on the other hand, holds potential to implement efficiently a large amount of vehicular applications thanks to its property of geographic content replication, while Software Defined Networking (SDN), is an attractive solution for the lack of flexibility and dynamic programmability that characterizes current VANET architectures. By implementing a logical centralization of the network, SDN enables dynamic and efficient management of network resources. In this paper, for a few reference scenarios, we illustrate how approaches that combine CCN, FC and SDN enable an innovative adaptive VANET architecture able to efficiently accommodate to intermittent connectivity, fluctuating node density and mobility patterns on one side and application performance and network resources on the other side, aiming to achieve high QoS. For each scenario, we highlight the main open research challenges, and we describe possible solutions to improve content dissemination and reduce replication without affecting content availability.

On the Flexibility and Autonomy of 5G Wireless Networks

With the emergence of the fifth generation (5G) wireless networks, not only is the increase in mobile broadband targeted, but also the support of various novel use cases, such as industrial automation, autonomous vehicles, e-health, and Internet of Things together with their requirements leading to highly heterogeneous wireless networks. This requires a re-design of the network architecture to ensure the coexistence of these use cases and guarantee user experience and service requirements. Therefore, 5G networks will be highly flexible and support online learning and autonomous decision making capabilities in a centralized and distributed manner to ensure highly efficient management of wireless and network resources. In this paper, the main features enabling flexibility and autonomy in 5G networks are discussed together with potential applications in different layers of the wireless network.

A time-domain control signal detection technique for OFDM

Transmission of system-critical control information plays a key role in efficient management of limited wireless network resources and successful reception of payload data information. This paper uses an orthogonal frequency division multiplexing (OFDM) architecture to investigate the detection performance of a time-domain approach used to detect deterministic control signalling information. It considers a type of control information chosen from a finite set of information, which is known at both transmitting and receiving wireless terminals. Unlike the maximum likelihood (ML) estimation method, which is often used, the time-domain detection technique requires no channel estimation and no pilots as it uses a form of time-domain correlation as the means of detection. Results show that when compared with the ML method, the time-domain approach improves detection performance even in the presence of synchronisation error caused by carrier frequency offset.