Edge server deployment is a critical component of mobile edge computing for meeting low-latency requirements while promoting sustainable and energy-efficient operation. Practical deployment is challenging because latency, energy consumption, and load balance are tightly coupled, and the decision space grows rapidly with network scale under time-varying traffic demand. Despite extensive studies, existing approaches often exhibit a scale–adaptivity tradeoff in large-scale scenarios: Scalable placement strategies are typically static under temporal demand variations, whereas adaptive learning-based methods may incur slow convergence or high training cost in high-dimensional deployment spaces. To address this, we propose SC-DQN, a scalable two-stage strategy integrating spectral clustering (SC) and deep Q-network (DQN). SC first partitions base stations using geographic proximity and connectivity information to reduce dimensionality and provide a structured initial deployment. DQN then iteratively refines server placement under dynamic traffic demand, where multiobjective optimization is implemented via a scalarized reward to explicitly control the latency–energy–load tradeoff. Extensive experiments on real-world Shanghai Telecom data demonstrate that SC-DQN achieves improved overall performance compared with representative baselines across multiple scenarios, improving load balancing and reducing latency and energy consumption by up to 28.47%, 34.82%, and 7.08%, respectively.

Emerging applications like smart cities necessitate rapid and localized data processing. To meet this critical requirement, the strategic deployment of edge servers within Mobile Edge Computing (MEC) architectures is essential to improve system performance and user experience. However, despite existing methods attempting to address the edge server deployment problem, challenges remain in large-scale deployment scenarios, including low deployment efficiency and reduced service quality. To address these issues, we propose MF-ESD, a novel edge server deployment strategy that leverages Mean Field Reinforcement Learning (MFRL) to optimize the deployment process. Specifically, we first employ an improved Whale Optimization Algorithm (IWOA) with adaptive weights and differential mutation for preprocessing to find a high-quality initial deployment scheme, thereby alleviating the cold-start problem in MFRL. Subsequently, MFRL utilizes the deployment results from IWOA and employs mean field approximation to determine the optimal edge server deployment strategy, which enhances the search efficiency. We validate the proposed method using real-world datasets provided by Shanghai Telecom and compare it against four baseline algorithms: Random, Top-K, Particle Swarm Optimization (PSO), and ESL. The experimental results show that MF-ESD reduces average latency and energy consumption while significantly improving load balancing performance, outperforming the baseline algorithms.

The distance vector hop algorithm is commonly used for sensor node localization. However, its high localization accuracy error and stability issues make it unsuitable for many applications. To overcome these concerns, this article proposes a new function and binary distance vector hop (FBDV-Hop) algorithm with binary controllers and function correction methods while considering the application requirement. In FBDV-Hop, binary controllers are designed to analyze the optimization effect of the module fully and make it adaptable to diverse scenarios. The correction strategies were based on average hop distance measurement, estimated distance, equation composition method, and localization after supplementary correction, which were divided into four modules in accordance with the module error sources of different design correction functions. The simulation experiment was designed to analyze the principle of the role of each module in depth and achieve the optimal optimization effect. The experimental results show that the localization error optimization rate under the FBDV-Hop algorithm was more than 70%, and the optimization rate, stability, effectiveness, and adaptability of the algorithm were considerably greater than the baseline algorithms.

This paper proposes an edge server deployment strategy based on multi-agent reinforcement learning (CKM-MAPPO) to address the multi-objective optimization problem in a vehicle networking environment. CKM-MAPPO focuses on optimizing the load balancing among edge servers and minimizing edge servers’ delay and energy consumption. Firstly, the Canopy and K-means algorithms determine the edge server deployment’s number and initial location. Then, we utilize the multi-agent reinforcement learning algorithm to decide the optimal deployment location of the edge server. We performed a series of tests, and the experimental results show that CKM-MAPPO improves load balancing by 26.5% and reduces delay and energy consumption by 12.4% and 17.9%, respectively.

Mobile edge computing (MEC) enables real-time processing and reduces core network congestion by bringing computing resources closer to data sources. However, optimizing edge server deployments to minimize latency, energy consumption, and load imbalance remains challenging. We propose the SC-DQN strategy, combining spectral clustering and deep Q-network (DQN) techniques. Spectral clustering analyzes the spatial distribution of base stations, grouping them and identifying cluster centers as initial deployment sites. A deep reinforcement learning environment then enables each edge server (agent) to iteratively optimize deployment, minimizing latency, energy consumption, and load imbalance. Experiments with Shanghai Telecom data demonstrate that SC-DQN improves load balancing by 23.05%, reduces latency by 45.32%, and lowers energy consumption by 9.64%, outperforming traditional methods.

To solve the problems of slow convergence and limited prediction ability of the Differential Evolution (DE) algorithm, an enhanced DE algorithm with the trustworthiness of a dynamic-selection framework (denoted by DEDF) is proposed. DEDF develops a trusted framework containing five mutation strategies to realize the dynamic selection of mutation strategies. On the basis of the framework, the mutation factor, crossover factor, and local exit strategy are improved to balance the algorithm’s local search and global search ability. A series of tests on the function set CEC2017 has been performed, and the findings show that compared with other benchmark algorithms, the DEDF has advantages in convergence speed and accuracy. The proposed algorithm DEDF can be effectively leveraged to address multi-objective optimization issues in IIOT.

In the Internet of Vehicles, low service delay and fast response are two essential factors to ensure the safety and smooth operation of vehicle networking. As core technologies, the 5G and edge computing networks play a fundamental role in reducing the pressure on the backbone network of vehicle networking and decreasing the service exchange delay. The previous edge server placement strategy can not be directly applied to deploying vehicle networking services, resulting in the degradation of system performance and user quality of experience. To solve the above problem, we propose a reliable edge server deployment algorithm called CFD based on Deep Deterministic Policy Gradient (DDPG). Firstly, the Canopy algorithm is used to cluster the location information of roadside units, and the initial cluster number is obtained. Then, the fuzzy C clustering algorithm (FCM) is leveraged to remove the "noise" and acquire the roadside units’ initial division and priority matrix. Finally, based on the DDPG algorithm, the optimal division of roadside units is obtained, and the cluster center is utilized as the deployment location of the edge server. Many experiments have been conducted, and the results show that, compared with the benchmark algorithm, the CFD algorithm improves the load balancing degree by 25%.

Differential evolution (DE) algorithm can be used in edge/cloud cyberspace to find an optimal solution due to its effectiveness and robustness. With the rapid increase of the mobile traffic data and resources in a cybertwin-driven 6G network, the DE algorithm faces some problems such as premature convergence and search stagnation. To deal with the problems mentioned above, in this article, an improved DE algorithm based on hierarchical multistrategy in a cybertwin-driven 6G network (denoted by DEHM) is proposed. Based on the fitness value of the population, DEHM classifies the population into three sub-population. Regarding each sub-population, DEHM adopts different mutation strategies to achieve a tradeoff between convergence speed and population diversity. In addition, a new selection strategy is presented to ensure that the potential individual with good genes is not lost. Experimental results suggest that the DEHM algorithm surpasses other benchmark algorithms in the field of convergence speed and accuracy. The proposed DEHM is expected to be leveraged in edge/cloud cyberspace, aiming at reducing energy costs and improving resource utilization.

The high computational capability provided by a data center makes it possible to solve complex manufacturing issues and carry out large-scale collaborative cloud manufacturing. Accurately, real-time estimation of the power required by a data center can help resource providers predict the total power consumption and improve resource utilization. To enhance the accuracy of server power models, we propose a real-time energy consumption prediction method called IECL that combines the support vector machine, random forest, and grid search algorithms. The random forest algorithm is used to screen the input parameters of the model, while the grid search method is used to optimize the hyperparameters. The error confidence interval is also leveraged to describe the uncertainty in the energy consumption by the server. Our experimental results suggest that the average absolute error for different workloads is less than 1.4% with benchmark models.

Virtualized networked datacenters (VNDCs) are gaining considerable attention for stochastic task execution under real-time constraints. However, the problem of efficiently minimizing the high energy consumption while ensuring high quality of service (QoS) in VNDCs has not been fully addressed. Although many solutions have been proposed to address this challenge, they are not efficient and only consider one or two of the energy consuming resources of VNDCs. To this end, an adaptive energy-aware algorithm,MCEC, that efficiently reduces the energy consumption of VNDCs while ensuring high QoS is proposed. Different from the existing approaches, the MCEC algorithm considers energy consumed by computing resources, virtual machine (VM) reconfiguration, communication resources and storage media resources while meeting user QoS requirements defined in the service level agreement (SLA). To validate the effectiveness of our algorithm, we carried out extensive experiments and compared the performance of our algorithm with existing baseline algorithms. The results of the experiments show that our algorithm substantially outperforms the baseline algorithms with respect to reducing energy consumption while respecting the service level agreement.

Resource optimization algorithm based on clustering and improved differential evolution strategy, as a new global optimized algorithm, has wide applications in language translation, language processing, document understanding, cloud computing, and edge computing due to high efficiency. With the development of deep learning technology and the rise of big data, the resource optimization algorithm encounters a series of challenges, such as the workload imbalance and low resource utilization. To address the preceding problems, this study proposes a novel resource optimization algorithm based on clustering and an improved differential evolution strategy (Multi-objective Task Scheduling Strategy (MTSS)). Three indexes, namely task completion time, execution cost, and workload, of virtual machines are selected and used to build the fitness function of the MTSS algorithm. At the same time, the preprocessing state is set up to cluster according to the resource and task characteristics to reduce the magnitude of their matching scale. Moreover, to solve the workload imbalance among different resource sets, local resource tasks are reallocated using the Q-value method in the MTSS strategy to achieve workload balance of global resources and improve the resource utilization rate. Experiments are carried out to evaluate the effectiveness of the proposed algorithm. Results show that the proposed algorithm outperforms other algorithms in terms of task completion time, execution cost, and workload balancing.

Summary With the increasing scale of tasks in cloud computing, the problem of high energy consumption becomes increasingly serious. To deal with the problem, we propose a cloud computing energy consumption model, which takes into account the execution and transmission cost of the processor. Then, based on this model, we put forward a task scheduling optimization algorithm named modified particle swarm optimization (M‐PSO) to handle the local optimum and slow convergence problem. Different from the PSO, M‐PSO can dynamically adjust the inertia weight coefficient to improve the speed of convergence according to the number of iterations. Finally, the performance of the proposed algorithm is evaluated through the CloudSim toolkit, and the experimental results show that the M‐PSO can efficiently reduce total cost compared with other algorithms.

Summary Cloud computing has gained more and more attention from industrial and academic circle since it offers pay‐as‐you‐go model, and business applications based on the cloud are also increasing. These applications meet the requirement of users while at the same time triggering the problem of high energy consumption in data centers. To deal with the problem, we propose a new algorithm named EEOM (Energy Efficiency Optimization of VM Migrations). Under considering CPU and memory factors, the key three steps for EEOM algorithm, including trigger time, VM selection, and host location, are optimized. EEOM algorithm takes use of the virtualization technology and migrates some VMs on the lightly loaded host and heavily loaded host to other hosts. The idle hosts are switched to low‐power mode or shut down so as to save energy consumption. The experimental results show that, as compared with Double Threshold (DT) algorithm, the EEOM algorithm saves 7% energy consumption and reduces 13% SLA violations.