The softwarization of networks is increasingly spreading, and one of the main paradigms is SDN (Software-Defined Networking), which allows overcoming the limitations mainly arising from the integration of the control plane and the forwarding plane within the network devices. It extracts the control plane to place it within a new logically centralized component: the SDN controller. Since this is a monolithic architecture that limits reliability and scalability, distributed solutions based on microservices have been proposed in the literature. In parallel, Agents are fully intelligent, atomic, and autonomous decision-making units that can be flexibly recomposed to create a completely autonomous network system. They also have the ability to replicate single or multiple decision-making processes that collaborate with each other. The development of future networks such as 5G, including 6G, is pushing towards the concept of network management automation and integration of intelligence, making agents an excellent means to meet this trend. This paper first introduces intelligence in the form of agents to a distributed SDN controller based on microservices, by implementing two new functionalities: topology _learning and shortest path, Then, it leverages a microservices-based SDN solution based on Ryu SDN framework, named MSN, to run agents in a Docker Container environment. Multiple measurements were performed locally in a single machine. Results show the topology learning performances compared with several network topologies. Moreover, the shortest patti agent experimental evaluations show the knowledge size depends on the network topology and the performances of different algorithms.

The emergence of 6G connectivity heralds a transformative era in wireless communication, emphasizing the necessity of network-wide intelligence for fully automated operations. At the heart of this paradigm shift lies the crucial need for an efficient algorithm capable of accurately predicting network traffic dynamics. This paper presents a comprehensive comparative study between two pivotal neural network architectures, LSTM (Long Short-Term Memory), and SP-LSTM (Speed-Optimized LSTM), within the context of network traffic prediction for the burgeoning 6G landscape. While LSTM stands as a widely acknowledged recurrent neural network, our innovative SP-LSTM is meticulously engineered for rapid and efficient decision-making. Through rigorous evaluation in a controlled environment, both models are examined for their efficacy in forecasting network traffic patterns. LSTM's proficiency in capturing long-term dependencies in sequential data is juxtaposed against SP-LSTM's emphasis on delivering swift and precise predictions. Our comparative analysis elucidates the distinctive strengths of these algorithms, assessing LSTM's effectiveness under varying conditions and scrutinizing SP-LSTM's proficiency in providing rapid, accurate forecasts. Additionally, we provide a GitHub link for accessing the project. This study offers vital insights into the relative merits and limitations of LSTM and SP-LSTM in network traffic prediction, essential for the advancement of intelligent 6G networking. These insights inform the selection of the optimal algorithm for realizing a seamless, intelligent 6G future with unparalleled capabilities.

Microservice architecture is a service-oriented paradigm that enables the decomposition of cumbersome monolithic-based software systems. Using microservice design principles, it is possible to develop flexible, scalable, reusable, and loosely coupled software that could be containerized and deployed in a distributed edge/cloud environment. The flexible deployment of microservices in an edge environment increases system performance in terms due to dynamic service function placement and chaining possibly resulting in latency reduction, fault tolerance, scalability, efficient resource utilization, cost reduction, and energy consumption reduction. On the other hand, virtualization and containerization of microservices add processing and communication overheads. Therefore, to evaluate end-to-end microservices-based system performance, we need to have an end-to-end mathematical formulation of the overall microservice-based network system. Incorporating the virtualization overhead, here we provide end-to-end mathematical formulation considering system parameters: latency, throughput, computational resource usage, and energy consumption. We then evaluate the formulation in a testbed environment with the Microservice-based SDN (MSN) framework that decomposes the Software-defined Networking (SDN) controller in microservices with Docker Container. The final result validates the presented mathematical modeling of the system’s dynamic behavior which can be used to design a microservice-based system.

Future networks such as 6G are projected to incorporate in-network intelligence toward achieving a zero-touch network. In this regard, several approaches proposed for the organizational architecture of future networks. Similar to microservice-based service design, a multi-agent-based network automation architecture was proposed as a competing paradigm for service-oriented architecture. The proposed architecture outlines the design guideline for intelligent network systems using agents as atomic and autonomous service units that can be used as building blocks. As a continuation of this approach, we design a Quality of Service (QoS) agent to control and manage stringent services such as remote surgery. QoS agent is intelligent that can capture and respond proactively to network traffic and workload distribution, showing hourly and seasonal patterns. QoS agents can be used as a building block along with traffic classification and traffic prediction agents for an intelligent networking system. For evaluation, a campus network is designed using a NetSim environment considering a three-tier network architecture. The QoS agent dynamically finds the best path for a particular service depending on the requirements. The agent communicates with the traffic prediction agent and traffic classifier agent to collect information about the network and services. Moreover, the QoS agent also observes the network states. Using these values along with existing network topology knowledge, it ranks the available paths to proactively allocate the best path for a service. Evaluation results suggest that with the appropriate accuracy of traffic prediction, the proposed approach can autonomously adapt in allocating a path for a given service.

Periodic Automatic Replenishment (PAR) is an inventory management policy that assists the healthcare sector in keeping the right amount of stock on hand to avoid excess stock and the potential for products to expire. Traditionally, hospitals rely on the experience and firsthand knowledge of stock management technicians to keep their store supplies with enough equipment. However, manual management based on “gut feeling“ and/or nursing feedback may lead to missing products or incorrect stock orders. Extracting accurate data is often too complex or time-consuming, resulting in a lack of critical reporting data to manage product inventories across the organization properly. However, adopting forecasting techniques and incorporating them into traditional PAR management policies can provide efficient solutions for controlling hospital warehouse inventory at the lowest cost. We have proposed a deep learning-based framework to monitor inventories to reduce costs and variation, create efficiencies, and improve the quality of patient care in hospitals. Furthermore, the proposed forecasting framework's performance is assessed using a real-world scenario within a hospital. The findings indicate that the system is capable of accurately tracking the usage of medical equipment, which results in a significant reduction in the unavailability of assets.

6G is aiming at fully incorporating in-network intelligence towards automated network management. In this regard, a multi-agent-based network automation architecture as a service design is proposed. The architecture introduces in-network intelligence, designing intelligent agents as the fundamental unit which is used as a building block in autonomous network system design. This work focuses on the dynamic agent placement problems in edge/cloud data centers. Agents are softwarized and intelligent versions of network functions that are traditionally implemented in hardware such as firewalls, packet gateways, etc. This paper, based on a combination of proactive and reactive solutions, considered decision accuracy in developing an intelligent decision algorithm that can be used in the prediction agent design. The proactive decision is based on a deep learning prediction algorithm using time-series workload forecasting. However, in case of unforeseen events that are missing from the historical dataset, the proactive decisions could be less reliable. Therefore, network-state feedback should be considered to determine the current network conditions using the change in instant arrival rate as a reactive decision. Then combining it with the proactive decision should be able to capture the unpredictable traffic spikes. The result is used to determine the number and type of agents to instantiate at a given time in the edge/cloud data centers. Using a public dataset in our algorithms, we predicted the workload request for a few days. The result shows improved decision accuracy over the existing solutions using the appropriate amount of dataset, machine learning models, and rate estimation.

Network softwarization, containerization, and cloudification in a distributed and centralized environment are the current tread in 5G, Beyond 5G, and 6G. In that sense, significant activities are going on in the research community to softwarize the network functions deploying them in a cloud-native environment. Cloud-native architecture is an approach for network function and service to be built specifically to deployed in the cloud. In this paper, we emulated Long LTE/LTE-A/5G in lightweight containers. We have evaluated the feasibility of using very lightweight environments such as k3s. We show the possibility of emulating a simple 4G/5G scenario without requiring a full Kubernetes infrastructure. The final results are based on available open projects, used as inspiration and as a starting point to modify deployment techniques and configurations. We have also explored scalability, orchestration, automation, and reliability of the deployments. Finally, we measured and tested the performance. The performance evaluation shows the potential application of the open platform-based emulations and deployment in 5G and beyond.

An autonomic network management system (ANMS) is expected to play a significant role in fifth and sixth-generation (5G and 6G) networks. It enables the network to manage itself with minimum or no human intervention. Recently, an ANMS architecture called multi-agent-based network automation of the network management system (MANA-NMS) architecture was presented. The article discussed a multi-agent service decomposition architecture, defining atomic network-functions (ANFs). These ANFs are proposed to be intelligent and autonomous agents. The agents are designed as independent atomic decision elements incorporating machine learning (ML) as an internal cognitive component. The atomic units are used as a building block for an ANMS. In line with this approach, this article proposes a network traffic classifier agent (NTCA) as a part of the network traffic management system. We first design and implement a NTCA using an ML algorithm as a cognitive component of the agent. To compare, we used K-Nearest Neighbors (K-NN), Decision Tree, Support Vector Machine (SVM), and Naive Bayes in the agent design. We perform an evaluation using classification accuracy, training latency, and classification latency. Finally, we tested the performance of the NTCA by implementing it in the MANA-NMS conceptual framework. The results show that the Decision Tree NTCA has the highest mean classification accuracy, the least mean training latency, and the lowest mean classification latency.

Internet of Things (IoT) network is dominating both the research and industry. There are numerous emerging IoT connectivity Technologies such as Sigfox, LoRa, NB-IoT, LTEM. However, these IoT connectivity technologies have different protocols and packet/message formatting. Thus, IoT devices are usually not able to interact with one another, causing interoper-ability challenges. This is creating the so-called network island or silos. Interoperability between different IoT networks needs to be achieved to fully exploit IoT potential. This is required at each level of the network. Different solutions have been proposed to tackle the interoperability problem at different levels reducing the difficulty in defining a solution breaching the vertical silos barrier. In this article, we focus on addressing network-level interoperability. We provide a network format translator in a virtualized environment as a flexible and lightweight deployment. As a proof of concept, a testbed is developed implementing the proposed translator using NS3. Using the testbed, we can communicate with different IoT technologies sending packets between each device in each type of IoT network. For example, sending a LoRaWAN packet to Wi-Fi and 6LoWPAN and visa-versa. Finally, we have measured the latency introduced by the translator.

The introduction of the Internet of Things (IoT) and massive machine-type communications has implied an increase in network size and complexity. In particular, there is already a huge number of IoT devices in the market in various sectors, such as smart agriculture, smart city, smart home, smart transportation, etc. The IoT interconnectivity technologies are also increasing. Therefore, these are increasingly overwhelming the efforts of network administrators as they try to design, reconfigure and manage such networks. Relying on humans to manage such complex and dynamic networks is becoming unsustainable. Network automation promises to reduce the cost of administration and maintenance of network infrastructure, by offering networks the capability to manage themselves. Network automation is the ability of the network to manage itself. Various standardization organizations are taking the initiative in introducing network automation, such as European Telecommunication Standardization Institute (ETSI). ETSI is leading the standardization activities for network automation. It has provided different versions of reference architecture called generic autonomic network architecture (GANA), which describes a four-level abstraction for network-management decision elements (DEs), protocol level, function level, node level, and network level. In this article, we review and survey the existing works before and after the introduction of software-defined networking (SDN) and network-function-virtualization (NFV). We relate the main trending paradigms being followed, such as SDN, NFV, machine learning (ML), microservices, multiagent system (MAS), containerization, and cloudification, as a pivotal enabler of full network automation. We also discuss the autonomic architectures proposed in the literature. Finally, we presented possible future research directions and challenges that need to be tackled to progress in achieving full network automation.

The advent of network softwarization is enabling multiple innovative solutions through software-defined networking (SDN) and network function virtualization (NFV). Specifically, network softwarization paves the way for autonomic and intelligent networking, which has gained popularity in the research community. Along with the arrival of 5G and beyond, which interconnects billions of devices, the complexity of network management is significantly increasing both investments and operational costs. Autonomic networking is the creation of self-organizing, self-managing, and self-protecting networks, to afford the network management complexes and heterogeneous networks. To achieve full network automation, various aspects of networking need to be addressed. So, this article proposes a novel architecture for the multi-agent-based network automation of the network management system (MANA-NMS). The architecture rely on network function atomization, which defines atomic decision-making units. Such units could represent virtual network functions. These atomic units are autonomous and adaptive. First, the article presents a theoretical discussion of the challenges arisen by automating the decision-making process. Next, the proposed multi-agent system is presented along with its mathematical modeling. Finally, MANA-NMS architecture is mathematically evaluated from functionality, reliability, latency, and resource consumption performance perspectives.

5G and beyond need to meet stringent requirements of latency, reliability, and support for heterogeneous devices. However, the existing wireless network architecture is limited to fulfill these constraints. Cloud radio access network, along with network function virtualization, is suggested to provide flexibility and network agility. It decouples network functions, such as firewall and packet gateway, from hardware to software deployed in the cloud. Thus comprehensive end-to-end formulation of this architecture is required for virtual network function placement. Most of existing works focus on virtual functions placement with different objectives, addressing different service requirements separately. In this article, six 5G constraints are considered simultaneously to find optimal virtual network function placement with service differentiation. The selected six parameters reflect services' requirements, network constraints and computing constraints. We first model the overall cloud radio access network as a multi-layer loopless-random hypergraph and we provide the overall formulation of the system. Then, we reformulate such model considering backup virtual functions and CPU over-provisioning techniques to improve both virtual function's reliability and processing latency. Finally, we propose service differentiation to reduce CPU utilization and energy consumption, while using the above techniques. The results suggest that the application of service differentiation can significantly improve assignment of computing resources and energy efficiency.

The energy consumption in data centers is drastically increasing and becoming a significant portion in the data center operating expenses. Enabling a sleep mode in the idle computing servers and network hardware is the most efficient method to avoid unnecessary power consumption. However, changes in the power modes introduce considerable delays. Moreover, inability to wake up a sleeping server immediately requires an availability of a pool of idle servers able to accommodate incoming load in the short term to prevent QoS degradation. In this paper we investigate the amount of computing servers and network hardware needed to accommodate different incoming load patters in the data centers. Furthermore, we propose to build these servers on energy efficient hardware, which is costly but can scale its power consumption with the offered load levels. The evaluation results show that the proposed methodology can save up to $750 per server per year on average.