Encyclopedia
Please note this is an old version of this entry, which may differ significantly from the current revision.
The widespread use of technology has made communication technology an indispensable part of daily life. However, the present cloud infrastructure is insufficient to meet the industry’s growing demands, and multi-access edge computing (MEC) has emerged as a solution by providing real-time computation closer to the data source. Effective management of MEC is essential for providing high-quality services, and proactive self-healing is a promising approach that anticipates and executes remedial operations before faults occur. The term self-healing (SH) is closely associated with autonomous computing (AC), which was introduced at IBM’s event in 2001. The event aimed to develop a system that could manage itself without human intervention to address the management issues arising from the growing computer systems and networks.
- computingcomputing
- cyber
- networks
1. Mobile Network Management with Autonomous Approaches
The heterogeneous nature of mobile edge computing has presented challenges in current network management approaches [50]. The complexity and diversity of the infrastructure and services provided make management increasingly difficult. As a result, there is a growing need for a network management approach that can consistently deliver high-quality services.
Traditionally, mobile networks rely on human experts to manually correct network faults. This reactive approach is inefficient and can negatively impact the quality of the experience. To address this, there has been a push for autonomous network management approaches that can automatically detect and remediate faults without human intervention.
The critical function of an autonomous system is its ability to operate without human intervention. An autonomous system (AS) is capable of performing computations on its own with little or no human involvement, as it is an evolving concept that is applicable in various fields, including maritime, space exploration, military, computing, manufacturing, robotics, and navigation [51].
An autonomous system (AS) is characterized by its hardware and software capabilities, which work together to identify a problem and execute a solution. It should also be able to gather information from external sources, process it, and make decisions accordingly, as it is a concept applicable in various fields, such as cloud computing, edge computing, and network management, which employ different approaches to achieve autonomous functionalities.
In computing technologies, such as cloud and edge computing, several methods are used to implement autonomous functionalities. Autonomous computing is one such approach. In network management, a self-organizing network (SON) is an approach that can be utilized to achieve autonomous functionalities.
2. Autonomous Approaches in MEC and Related Technologies (IoT and Bigdata)
Mobile edge computing is a critical enabler of IoT and big data technologies. By bringing computation and storage closer to the data sources, MEC accelerates the process of IoT data sharing and analytics. The autonomous decision-making principle is vital to enhancing the management of these technologies and services. It provides for real-time management without human intervention, making it essential for the Internet of Things (IoT), 5G, big data, mobile networks, and other related technologies.
IoT consists of various sensor devices, and the autonomic decision-making concept is essential for the installation and management of these devices. Manual consideration is ineffective, due to the large number of devices involved, making a dynamic and intelligent management technique necessary.
The autonomic control loop, consisting of autonomic managers and managed resources, can be introduced to the IoT component structure as long as the hierarchy remains. The components above serve as the autonomic manager, and the components below serve as managed resources. This approach can be applied to the components individually or collectively.
Autonomic decision making is critical for the effective management of IoT and big data technologies, which heavily rely on mobile edge computing to bring computation and storage closer to the data source. The real-time management provided by autonomic decision making without human intervention is particularly important in device management, access management, and identity management. By introducing the autonomic decision-making concept, the system can control and manage its functions automatically, resulting in greater effectiveness in all components.
Research initiatives have explored the fusion of IoT and autonomous control systems to create a responsive environment capable of adapting to changes. Lei et al. [52] proposed a system that utilizes sensors to gather data, servers to make decisions based on data analysis, and actuators to carry out those decisions. To achieve autonomy, the authors use reinforcement learning and deep reinforcement learning in the decision-making process. However, utilizing deep reinforcement learning for decision making can be affected by delayed control and incomplete perception.
To achieve autonomous functionalities in IoT, Wei et al. [53] proposed the use of broad reinforcement learning (BRL) to reduce traffic jams in smart cities. The proposed approach by Wei et al. highlighted the growing interest in using autonomous decision making in IoT management. By leveraging BRL, their solution aimed to reduce traffic congestion in smart cities by controlling traffic lights. The system’s reinforcement learning algorithm learns the optimal timing and sequence of traffic lights based on real-time traffic data.
This approach demonstrates the potential of autonomous decision making in not only managing devices, but also in optimizing complex systems. With the increasing prevalence of IoT devices in smart cities, there is a growing need for autonomous management to ensure the efficient operation of these systems. Autonomous decision-making approaches, such as BRL, can be used in various applications, such as energy management, waste management, and public safety.
However, as with any autonomous system, there are also potential risks and challenges that need to be considered. The algorithm used in Wei et al.’s approach relies hvily on real-time traffic dataea. This raises questions about data privacy and security. In addition, there is a risk of the system being manipulated by malicious actors, which could lead to traffic accidents or other safety concerns.
To ensure the safe and effective implementation of autonomous decision making in IoT management, it is essential to address these potential risks and challenges. This includes developing robust security measures, ensuring data privacy, and implementing fail-safe mechanisms to prevent system failures. As research in this area continues to evolve, it will be crucial to strike a balance between the potential benefits and potential risks associated with autonomous decision making in IoT management.
Tahir et al. [54] have also proposed the use of autonomic computing for managing the IoT ecosystem with minimal human intervention. They describe the use of the MAPE (monitor-analyze-plan-execute) control loop for managing IoT components. The MAPE control loop assigns autonomic managers and managed resources to these components, enabling them to perform self-management and self-optimization without human intervention. This approach is particularly useful in scenarios where the number of devices and sensors is too large for manual management, and the system needs to adapt to changing conditions in real-time.
Big data is a field where autonomous functionalities are increasingly being applied to achieve greater efficiency in various tasks, including data security and privacy. One way in which autonomous decision making can improve security is through the automatic detection and prevention of intrusion. Traditional intrusion detection methods rely on sets of rules to analyze the source and behavior of an intrusion and apply preventive action, which can be slow and inefficient. This approach can result in a time gap between the detection and response, which can be crucial in preventing intrusions.
To overcome these limitations, there is a need to allow the system to protect itself without human intervention. This is where autonomous decision making can be applied to enhance the process of data security and privacy. By using machine learning algorithms to analyze data in real-time, an autonomous system can automatically detect potential intrusions and take preventive action, thereby reducing the response time and improving the overall effectiveness of the system. This approach is more efficient and reliable than traditional intrusion detection methods, making it a valuable tool for data security and privacy in the field of big data.
A number of studies have explored the use of autonomous functionalities in big data applications. One such study is that of Vieira et al. [55], which investigates the potential of integrating autonomic computing and big data techniques to build a system capable of processing large volumes of data from system operations to identify anomalies and formulate countermeasures in the event of an attack. The researchers propose a framework that uses machine learning algorithms to analyze data in real-time and make automatic decisions to protect the system. The framework is designed to be adaptive, so it can learn from new data and adjust its response accordingly. By automating the process of detecting and responding to threats, this framework provides a more effective way to secure big data systems, compared to traditional methods that rely on human intervention.
The autonomy intrusion detection system (AIRS) is a proposed method that is designed based on the MAPE-K model. One of its unique features is its ability to identify trends in massive log databases, which enables it to quickly detect and respond to cyber attacks. This is achieved by comparing signatures to suspicious actions, thus reducing the time lag between intrusion detection and response. The system’s efficient response time makes it an effective tool for improving data security and preventing cyber attacks.
Kassimi et al. [56] also explored the use of autonomous systems for intrusion detection, specifically the discovery of anomaly data. Their system is designed to gather information about network traffic through a mobile agent, which then stores and analyzes the data. Using Hadoop infrastructure, the large amount of data collected is organized and analyzed by multi-agent systems to provide a self-healing intrusion detection system. This autonomous approach to intrusion detection allows for the efficient and effective identification of security threats, without relying on human intervention, thus reducing response time and improving the overall security of the system.
In a related research effort, Mokhtari et al. [57] developed an autonomous proactive task dropping mechanism that utilizes probabilistic analysis to achieve autonomous functionality in a distributed computing system. The main aim is to maintain the system’s performance when faced with adverse situations. The authors employed a mathematical model to identify the most optimal task dropping decision that will enhance the system’s stability, followed by a task dropping heuristic to achieve stability within a reasonable time frame. By adopting this approach, the system can continue to operate without human intervention and quickly adapt to changes in the environment, ensuring that it remains stable and performs optimally.
3. Review of Autonomous Architecture and Framework for MEC
The concept of autonomy enables a system to coordinate itself without external intervention, which can be particularly useful in a MEC environment. In this subsection, we will introduce several autonomous architectures and frameworks found in the literature that can be applied in MEC environments.
Biologically inspired frameworks and architectures—the development of autonomic computing has been influenced by the human nervous system. The human nervous system has the ability to control and regulate the body’s systems, and this ability has been replicated in various autonomic computing approaches. These approaches focus on an entity that is responsible for controlling the systems, similar to how the brain controls the human body. By replicating the functions of the nervous system, autonomic computing can achieve self-regulation and self-optimization without external intervention.
There are various architectures and frameworks in the literature that can be useful in the context of MEC environments. Two of the relevant ones are biologically inspired frameworks and large-scale distributed system architecture.
Biologically inspired frameworks and architectures aim to replicate the human nervous system, which has the ability to regulate and control body systems without external intervention. In the context of AC, the approach is to create an entity responsible for controlling the system, similar to how the brain controls the human body.
Large-scale distributed system architecture includes applications such as Ocean store SMARTS from IBM and Auto Admin from Microsoft, which have been developed to manage distributed systems and databases automatically. These applications are designed to achieve the goals of AC, such as self-management and self-healing.
Another architecture that has been widely used to achieve autonomous behavior is the agent-based architecture. In this type of architecture, agents interact with each other to create a system that can oversee itself automatically. The key aspect of this architecture is the agent, which is an autonomous entity that can function without the need for a central control system.
The component-based architecture places emphasis on autonomic components and their interactions, with the goal of achieving a set objective. These components carry out monitoring, analysis, planning, and execution, which are the core functionalities of autonomous computing. By breaking down the system into smaller, more manageable components, this architecture enables a more efficient and effective approach to achieving autonomy. The autonomic components interact with each other, exchanging information and coordinating their actions to achieve the overall objective.
Technique-focused architecture is an AC architecture that leverages AI and control theory to optimize system performance. Its primary objective is to generate actions, acquire feedback, manage actions by retrying or re-planning, and generate alternative approaches to achieve specified goals. The architecture comprises features that enable learning from the environment, taking action, receiving feedback, planning, and re-planning, thus achieving autonomous behavior in a system.
On the other hand, service-oriented architecture focuses on the interaction of services to achieve autonomous functionalities. Services are loosely coupled and operate independently, offering interoperability and scalability. This architecture enables the creation of self-managed systems through the dynamic configuration of services, allowing for fault tolerance and self-healing capabilities.
This entry is adapted from the peer-reviewed paper 10.3390/computers12030063
This entry is offline, you can click here to edit this entry!
