A Review of Reinforcement Learning Based Intelligent Optimization for Manufacturing Scheduling

Complex System Modeling and Simulation, Год журнала: 2021, Номер 1(4), С. 257 - 270

Опубликована: Дек. 1, 2021

As the critical component of manufacturing systems, production scheduling aims to optimize objectives in terms profit, efficiency, and energy consumption by reasonably determining main factors including processing path, machine assignment, execute time so on. Due large scale strongly coupled constraints nature, as well real-time solving requirement certain scenarios, it faces great challenges problems. With development learning, Reinforcement Learning (RL) has made breakthroughs a variety decision-making For problems, this paper we summarize designs state action, tease out RL-based algorithm for scheduling, review applications RL different types then discuss fusion modes reinforcement learning meta-heuristics. Finally, analyze existing problems current research, point future research direction significant contents promote optimization.

Язык: Английский

---

Effectively Detecting Operational Anomalies In Large-Scale IoT Data Infrastructures By Using A GAN-Based Predictive Model

The Computer Journal, Год журнала: 2022, Номер 65(11), С. 2909 - 2925

Опубликована: Авг. 24, 2022

Abstract Quality of data services is crucial for operational large-scale internet-of-things (IoT) research infrastructure, in particular when serving large amounts distributed users. Effectively detecting runtime anomalies and diagnosing their root cause helps to defend against adversarial attacks, thereby essentially boosting system security robustness the IoT infrastructure services. However, conventional anomaly detection methods are inadequate facing dynamic complexities these systems. In contrast, supervised machine learning unable exploit due unavailability labeled data. This paper leverages popular GAN-based generative models end-to-end one-class classification improve unsupervised detection. A novel heterogeneous BiGAN-based model Heterogeneous Temporal Anomaly-reconstruction GAN (HTA-GAN) proposed make better use a classifier scoring function. The Generator-Encoder-Discriminator BiGAN structure can lead practical score computation temporal feature capturing. We empirically compare approach with several state-of-the-art on real-world datasets, benchmarks synthetic datasets. results show that HTA-GAN outperforms its competitors demonstrates robustness.

Язык: Английский

---

HUNTER: AI based holistic resource management for sustainable cloud computing

Journal of Systems and Software, Год журнала: 2021, Номер 184, С. 111124 - 111124

Опубликована: Окт. 22, 2021

Язык: Английский

---

AI augmented Edge and Fog computing: Trends and challenges

Journal of Network and Computer Applications, Год журнала: 2023, Номер 216, С. 103648 - 103648

Опубликована: Май 4, 2023

In recent years, the landscape of computing paradigms has witnessed a gradual yet remarkable shift from monolithic to distributed and decentralized such as Internet Things (IoT), Edge, Fog, Cloud, Serverless. The frontiers these technologies have been boosted by manually encoded algorithms Artificial Intelligence (AI)-driven autonomous systems for optimum reliable management resources. Prior work focuses on improving existing using AI across wide range domains, efficient resource provisioning, application deployment, task placement, service management. This survey reviews evolution data-driven AI-augmented their impact systems. We demystify new techniques draw key insights in Fog Cloud management-related uses methods also look at how can innovate traditional applications enhanced Quality Service (QoS) presence continuum present latest trends areas optimizing models that are deployed or layout roadmap future research directions QoS optimization reliability. Finally, we discuss blue-sky ideas envision this an anchor point AI-driven

Язык: Английский

---

Resource Allocation With Workload-Time Windows for Cloud-Based Software Services: A Deep Reinforcement Learning Approach

IEEE Transactions on Cloud Computing, Год журнала: 2022, Номер 11(2), С. 1871 - 1885

Опубликована: Апрель 21, 2022

As the workloads and service requests in cloud computing environments change constantly, cloud-based software services need to adaptively allocate resources for ensuring Quality-of-Service (QoS) while reducing resource costs. However, it is very challenging achieve adaptive allocation with complex variable system states. Most of existing methods only consider current condition workloads, thus cannot well adapt real-world subject fluctuating workloads. To address this challenge, we propose a novel Deep Reinforcement learning based Allocation method workload-time Windows (DRAW) that considers both future process. Specifically, an original Q-Network (DQN) prediction model management operations trained on windows, which can be used predict appropriate under different Next, new feedback-control mechanism designed construct objective plan state through iterative execution operations. Extensive simulation results demonstrate accuracy generated by proposed DRAW reach 90.69%. Moreover, optimal/near-optimal performance outperform other classic 3 $\sim$ 13% scenarios.

Язык: Английский

---

AERF: Adaptive ensemble random fuzzy algorithm for anomaly detection in cloud computing

Computer Communications, Год журнала: 2023, Номер 200, С. 86 - 94

Опубликована: Янв. 9, 2023

Язык: Английский

---

Multi-Objective Prioritized Task Scheduler Using Improved Asynchronous Advantage Actor Critic (a3c) Algorithm in Multi Cloud Environment

IEEE Access, Год журнала: 2024, Номер 12, С. 11354 - 11377

Опубликована: Янв. 1, 2024

Task scheduling is a crucial challenge in cloud computing paradigm as variety of tasks with different runtime processing capacities generated from various heterogeneous devices are coming up to application console which effects system performance terms makespan, resource utilization, cost. Therefore, traditional algorithms may not adapt this efficiently. Many existing authors developed task schedulers by using metaheuristic approaches solve problem(TSP) get near optimal solutions but still TSP highly dynamic challenging scenario it NP hard problem. To tackle challenge, paper introduces multi objective prioritized scheduler improved asynchronous advantage actor critic(a3c) algorithm uses priorities based on length tasks, and VMs electricity unit cost environment. Scheduling process carried out two stages. In the first stage, all incoming VM calculated at manager level second Priorities fed (MOPTSA3C) generate decisions map effectively onto considering schedule cost, makespan available Extensive simulations conducted Cloudsim toolkit giving input trace fabricated data distributions real time worklogs HPC2N, NASA datasets scheduler. For evaluating efficacy proposed MOPTSA3C, compared against techniques i.e. DQN, A2C, MOABCQ. From results, evident that MOPTSA3C outperforms for reliability.

Язык: Английский

---

A new cloud-based method for composition of healthcare services using deep reinforcement learning and Kalman filtering

Computers in Biology and Medicine, Год журнала: 2024, Номер 172, С. 108152 - 108152

Опубликована: Фев. 13, 2024

Язык: Английский

---

Intelligent Resource Orchestration for 5G Edge Infrastructures

Future Internet, Год журнала: 2024, Номер 16(3), С. 103 - 103

Опубликована: Март 19, 2024

The adoption of edge infrastructure in 5G environments stands out as a transformative technology aimed at meeting the increasing demands latency-sensitive and data-intensive applications. This research paper presents comprehensive study on intelligent orchestration computing infrastructures. proposed Smart Edge-Cloud Management Architecture, built upon an OpenNebula foundation, incorporates ONEedge5G experimental component, which offers workload forecasting automation capabilities, for optimal allocation virtual resources across diverse locations. evaluated different models, based both traditional statistical techniques machine learning techniques, comparing their accuracy CPU usage prediction dataset machines (VMs). Additionally, integer linear programming formulation was to solve optimization problem mapping VMs physical servers distributed infrastructure. Different criteria such minimizing server usage, load balancing, reducing latency violations were considered, along with constraints. Comprehensive tests experiments conducted evaluate efficacy architecture.

Язык: Английский

---

WS‐DOA: Design of Hybrid Heuristic Algorithm for Deriving Multi‐Objective Function of Optimal Task Scheduling and VM Migration Over Cloud Sector

Transactions on Emerging Telecommunications Technologies, Год журнала: 2025, Номер 36(4)

Опубликована: Март 20, 2025

ABSTRACT Cloud‐based computing is an innovative model that utilizes a variety of self‐driving devices and adaptable structures. Efficient cloud relies on the critical step scheduling tasks. In order to decrease energy use increase service providers' profits by speeding up processing, task planning remains crucial. Scheduling tasks represents one crucial operations in cloud. The main challenge allocate complete suitable Virtual Machine (VM) while ensuring profitability. Various techniques ensure Quality Service (QoS), but as scaling increases, becomes more challenging. Hence, there need for enhanced scheduling. Previous studies did not cover VM migration, which effectively address resource utilization efficiency. An advanced deep learning with heuristic algorithm suggested improve process. This aims predict data assist migration through derivation multi‐objective function. Initially, are gathered from benchmark sources. Further, prediction carried out Multiscale Dilated Recurrent Neural Network (MDRNN). To derive function, Water Strider‐based Dingo Optimization Algorithm (WS‐DOA) proposed. Following prediction, performed WS‐DOA function considering constraints like cost, consumption, response time, security. Likewise, involves formulating objective WS‐DOA, make span cost. Finally, proposed examined using diverse metrics. On contrary, method evinces it acquires higher results migration.

Язык: Английский

---