Search

In the harsh mobile edge computing (HMEC) environment, there are many dynamic changes such as interference from noise, the impact of extreme environmental conditions, and the mobility of devices. It is a great challenge to the online realtime task offloading scheduling for delay-sensitive applications. However, the dynamic changes in HMEC environment have been ignored in almost all previous studies. Therefore, we propose the federated deep reinforcement learning-based online task offloading and resource allocation (FD-OTR) algorithm to address the task offloading in HMEC. Additionally, the FD-OTR algorithm performs resource allocation for offloaded tasks. The task offloading part of FD-OTR algorithm can be divided into two layers: the deep reinforcement learning (DRL) layer and the federated learning (FL) layer. The online algorithm in the DRL layer can adapt to the dynamic HMEC environment and make real-time task offloading decisions. In the FL layer, federated learning with low communication overhead is used for model aggregation to form a better global model. Resource allocation is done by using a new meta-heuristic algorithm: the Sparrow Search Algorithm (SSA). Finally, the simulation results demonstrate that the FD-OTR algorithm performs well in HMEC. The convergence speed of FD-OTR is three times faster than the centralized method. Compared to the baseline algorithms, FD-OTR reduces costs by 14.3%, 11.2% and 9.28%, respectively. [ABSTRACT FROM AUTHOR]

For the global-scale multimedia live streaming services, both of the cost-efficiency at the service provider side and the Quality of Experience (QoE) satisfaction at the viewer side need to be achieved. This is a difficult challenge because the request patterns of global live-streaming services are highly dynamic. In this paper, we solve this issue by cloud-based adaptive resource allocation. We first present a cloud-based multi-tier architecture, called MaaS (Media as a Service), which consists of four types of modules. The main issue that we focus on is the deployment of properly dimensioned MaaS modules in proper geographical regions. We take the QoE of the Dynamic Adaptive Streaming over HTTP viewers into account for this decision. We propose a combination of deep-learning based demand prediction scheme and a dynamic-programming based heuristic to make a good tradeoff between viewers' QoE and the cloud resource cost. Extensive evaluation shows that the proposed scheme clearly outperforms the existing schemes. [ABSTRACT FROM AUTHOR]

Blockchain, a promising technology, has been extensively applied in numerous fields, such as network security, finance, and medical care. However, due to the low power consumption and weak computing power of the mobile environment, the application of blockchain in this environment still faces many challenges. Therefore, edge computing has been introduced to improve the computing power of mobile devices and encourage more mobile edge devices to join the blockchain network. In this paper, we propose a double auction model to address the issue of edge computing resource allocation in blockchain networks. Based on this auction model, we first propose a truthful double auction mechanism based on breakeven (TDAMB) to determine matched pairs of edge computing service providers (ECSPs) and miners. Furthermore, to improve the system efficiency, we propose a double auction mechanism based on a critical value (DAMCV). We also theoretically analyze the individual rationality, budget balance and truthfulness of the proposed mechanisms. Extensive experiments show that TDAMB and DAMCV have good effects on edge computing resource allocation in blockchain networks. [ABSTRACT FROM AUTHOR]

Device-to-Device (D2D) communication is an emerging technology for beyond fifth-generation (B5G) networks to support significant features, such as spectrum efficiency, high data rate, and reduced power consumption. Efficient spectrum allocation and optimum power control in D2D networks is one of the major issues that arise due to dynamic traffic arrivals. This article proposes a novel approach to maximize the spectral efficiency of the D2D network. The formulated problem is a mixed-integer nonlinear programming problem (MINLP). However, due to its complexity, the global optimal solution is difficult to solve directly. A two-stage optimization algorithm is presented: optimal resource allocation algorithm (ORAA) by utilizing the concept of queue dynamics and optimal power control algorithm by adapting the Lyapunov stability method. The proposed method achieves higher spectral efficiency up to 21.6% and latency is minimized up to 39.89% over the benchmark schemes. The proposed method shall find immense use in smart traffic management to support Intelligent transportation systems. [ABSTRACT FROM AUTHOR]

One of the most well-known ways for cloud service providers (CSPs) to satisfy customers and reduce SLA violations is to offer resources through an auction market. In previous research based on classical games, the CSP has been considered as a player. A drawback of this type of modeling is that it is not scalable as the number of CSPs increases. In this paper, we use the combination of barter and evolutionary game theory to model the exchange of VM instances among CSPs. First, each CSP estimates the free and used resources. Then, according to the estimated valuation, the CSP announces its bid as a strategy to the auctioneer. We use evolutionary game theory to update the strategy of CSPs and increase their winning probability. The proposed method does not require any monetary exchange or registration of separate contracts between CSPs. In case of an SLA violation, a CSP can reclaim its rented resources from the other party. Moreover, since the violation of the agreement is costly for the violating party, the motivation for cooperation between CSPs increases. The simulation results show that the proposed model can lead to a 5% improvement in social welfare compared to the state-of-the-art barter double auction methods. Moreover, in addition to increasing the payoff of CSPs, it reduces the number of contracts by 3%. [ABSTRACT FROM AUTHOR]

The Internet of Things (IoT) technology is used to develop a wide range of applications and services, including intelligent healthcare systems and virtual reality applications. Low processing power limits IoT devices' capabilities. It's common practice to use cloud services to do operations that would otherwise require a user's device to be overloaded with data. High latency, high traffic, and high energy consumption remain, though. Given the above concerns, Fog Computing (FC) should be applied in the IoT to speed up time-sensitive data processing and management. In this study, a novel architecture for offloading jobs and allocating resources in the IoT is presented. Sensors, controllers, and FC servers are all part of the upgraded system. The second layer uses the subtask pool approach to offload work and the Moth-Flame Optimization (MFO) algorithm combined with Opposition-based Learning (OBL) to distribute resources. This combination is known as OBLMFO. A stack cache approach is used to complete resource allocation in the second layer to avoid system load imbalance. In addition, the second layer relies on the blockchain to guarantee the accuracy of transaction data. Another way to put it is that the proposed architecture utilizes blockchain advantages to optimize resource distribution in the IoT. The evaluation of the OBLMFO model was done through the Python 3.9 environment, which contains a large variety of distinct jobs. The results show that the OBLMFO model reduced the delay factor by 12.18% and the energy consumed by 6.22%. [ABSTRACT FROM AUTHOR]

Cloud computing is a dynamic paradigm which aims to provide on-demand computing resources to a user over the Internet. Most of the existing cloud providers use fixed pricing strategy. However, a few cloud service providers have recently priced their services using auction-style pricing. While cloud consumers take care of Quality of Service (QoS), their preferences for QoS differ from those of cloud providers. Consumers that use the cloud, for example, typically expect high availability, therefore, they need to bring availability, reliability, and scalability together to achieve true high availability. In contrast, high availability entails more expenditures in hardware and/or software resources, resulting in higher expenses for cloud providers. In this paper, we propose a novel auction strategy to include QoS parameters bundle into the price, allocating available cloud services between users, and establishing an auction strategy to enable revenue approximation. In order to accomplish this, we formulate the problem of allocating a cloud service coupled with bundle of QoS parameters to the winners as a multi-attribute auction problem. To address this problem, we present a novel multi-attribute auction pricing model with numerous consumers and a single cloud provider. We also introduce the new QoS Parameter Bundle Pricing Scheme that establishes the relation between a cloud service's bundle of QoS parameters and the estimated price of cloud resources, which can produce more revenue than a fixed pricing strategy. Last, we put forward three algorithms MAAPM (i.e., to solve a multi-attribute auction problem), FMAA (i.e., represents existing state-of-the-art auction mechanisms in the literature) and Fixed-Payment (i.e., represents the currently used fixed payment strategy). Extensive simulations have revealed that MAAPM outperforms FMAA and Fixed Payment in all the metrics except the execution time, also MAAPM results at growing revenues for cloud providers and decreasing expenditure for users. [ABSTRACT FROM AUTHOR]

Based on virtualization technologies, virtual machines (VMs) provide computing services and network resources for cloud users over the Internet. When cloud users use VMs for an extended period of time, requests generated by other cloud users are easily blocked. When cloud users no longer use VMs, requests' throughput will be decreased substantially. We propose a multi-input cloud resource allocation strategy with limited buffer and VM synchronization failure, which can improve the throughput of requests. A limited buffer is added to the system to reduce the possible blocking behaviors in the proposed strategy. Moreover, we assume that a physical machine failure will put all the VMs in failure states. The system recovery will take a period of time after the failure, which is called repair time. After the failure is repaired, the system continues to provide services for cloud users. A 2-dimensional Markov chain is established to measure the performance indexes of requests. In addition, we show the blocking rate, the loss rate, the throughput, the average quantity, and the average latency of requests by system state transition probability matrix and analyze their changing trends through numerical experiments. Finally, considering the balance between requests' blocking rate and requests' throughput, we construct a system profit function to ascertain the optimal number of request streams and obtain the maximal system profit. [ABSTRACT FROM AUTHOR]

Fog computing has become a state-of-art technology for cloud applications in collaboration with physical IOT devices on the edge of the network.We propose a Fog of Things (FOT) framework for optimized resource allocation to manage issues of over demanding and load balancing. The work discussed here is based on the residential demand side model for the areas that generate large number of requests per hour and resource allocation to these requests that require large amount of time and resources. In our FOT Framework, we have two main layers: fog layer and consumer layer. The fog layer performs the resource allocation in optimal time using Jaya optimization algorithm. The consumer layer makes decision for selection of particular appliances in residential buildings using a Multiple Knapsack algorithm. Both layers act as players following the extensive form of the game theory approach to share their moves to each other. [ABSTRACT FROM AUTHOR]

Traditional scheduling techniques are designed to reduce processing times while disregarding energy costs. One way of lowering energy usage is to implement scheduling strategies that distribute tasks to specified resources, which influence the processing time and power usage. Among the primary objectives to be achieved in cloud computing, power and energy consumption of the cloud environment have become issues due to ecological and economic reasons. Despite the existence of research efforts from the past pertaining to the same topic, an ideal solution to this problem has not yet been found. One of the main drawbacks of utilizing cloud computing is that the cloud data centers hosting cloud computing applications use higher volumes of energy, adding to the increased functioning cost and carbon footprint in the environment, which in turn increases the need for energy-efficient systems. In this research work, a hybrid optimization algorithm is presented, which is intended to minimize energy in the cloud computing environment. Thus, the crow search algorithm (CSA) and the sparrow search algorithm (SSA) are combined to obtain the proposed hybrid model. The hybrid approach achieves the optimal position in the shortest period of time and with the least amount of energy, load, and makespan, hence improving system performance. The experiments were conducted using three setups with different task sizes. The analysis while using the task size = 300 shows that the proposed method improved QoS, Resource Utilization (RU) and decreased makespan, energy usage, and load at a rate of 0.5073, 4.4035, 0.0331, and 0.0014, respectively. [ABSTRACT FROM AUTHOR]

Recently, there has been growing interest in distributed models for addressing issues related to Cloud computing environments, particularly resource allocation. This involves two main approaches: task scheduling, where the Cloud provider assigns tasks to Virtual Machines (VMs), and VM-to-Physical Machine mapping. These aspects are closely linked to the crucial issue of energy consumption in Cloud computing. A systematic and comprehensive review of the recent literature published between 2016 and 2023 was conducted to address the challenges and highlight the current state of research in this field. The review also highlights new opportunities for future research and guides for researchers to develop new contributions or improve upon existing ones. This work aims to help advance the state of resource allocation in Cloud computing environments. [ABSTRACT FROM AUTHOR]

A cloud data center provides various facilities such as storage, data accessibility, and running many specific applications on cloud resources. The unpredictable demand for service requests in cloud workloads affects the availability of resources during scheduling. It raises the issues of inaccurate workload prediction, lack of fulfillment in resource demands, load unbalancing, high power consumption due to heavy loads, and problems of under and overutilization of resources. Therefore, an efficient scheduling technique and an accurate forecasting model are needed to overcome these issues. Also, to deal with these challenges and provide optimal solutions, researchers must have a robust knowledge of cloud workloads, their types, issues, existing technologies, their advantages and disadvantages. However, previous research indicates limited systematic review studies exist for cloud workload applications with prediction-based scheduling techniques. Therefore, a survey is required that provides information related to cloud workload. To fulfill this requirement, the current study collects the related articles published in the past years. This paper is a systematic review study of prediction-based scheduling techniques that extract and evaluate data based on five criteria. It includes the datasets of different workload applications, resources, current prediction and scheduling techniques, and their related parameters. The survey is quite useful for academicians who want to select the problem and develop new techniques for issues related to cloud workload applications. It also gives an idea of existing approaches that are already implemented and employed. [ABSTRACT FROM AUTHOR]

We have designed and implemented a new data processing framework called "MeLoN" (Multi-tenant dEep Learning framework On yarN) which aims to effectively support distributed deep learning applications that can show another type of data-intensive workloads in the YARN-based Hadoop ecosystem. MeLoN is developed as one of Hadoop YARN applications so that it can transparently co-host existing deep learning applications with other data processing workflows. In this paper, we present comprehensive techniques that can effectively support multiple deep learning applications in a Hadoop YARN cluster by leveraging fine-grained GPU over-provisioning policy and a high-performance parallel file system for data staging which can improve the overall system throughput. Through our extensive experiments based on the representative deep learning workloads, we demonstrate that MeLoN can make an effective convergence of deep learning and the big data platform Hadoop by employing YARN-based resource allocation and execution mechanisms for running distributed deep learning applications. We believe that MeLoN can bring many additional interesting research issues including profiling of expected GPU memory usages of deep learning applications, supporting more complicated deep learning related jobs based on queuing systems which can ultimately contribute to a new data processing framework in the YARN-based Hadoop ecosystem. [ABSTRACT FROM AUTHOR]

Big Data applications require more energy consumption to process a massive volume of data in a heterogeneous environment. Moreover, reducing energy consumption in Big Data applications is an important research topic. It is one of the challenging issues to conserve energy with a deadline constraint in a heterogeneous environment. In this paper, we formulate scheduling the MapReduce jobs as a minimization problem by considering the decision variables with a user-specified deadline constraint. Further, a Learning Automata-based MapReduce Scheduling (LA-MRS) algorithm has been proposed to identify the resource allocation and save energy consumption of MapReduce tasks in a heterogeneous environment. We perform experimentation on the proposed LA-MRS algorithm using Hibench benchmark workloads such as Enhanced DFSIO, Nutch Indexing, k-mean Clustering and Hive Join. The experimentation illustrates that the proposed LA-MRS algorithm schedules the MapReduce task by saving around 25% of less energy consumed when compared to the existing algorithms. [ABSTRACT FROM AUTHOR]

Neighborhood selection is an important part of a multiobjective evolutionary algorithm based on decomposition (MOEA/D) because the impetus for population evolution mainly comes from its neighborhood. However, the fixed neighborhood size used in MOEA/D may deteriorate the performance of the algorithm due to an unreasonable allocation of computational resources. To further improve the performance of MOEA/D, this paper proposes a multiobjective decomposition evolutionary algorithm with optimal history-based neighborhood adaptation and a dual-indicator selection strategy. The optimal history-based neighborhood adaptation strategy is applied to alleviate the imbalance between exploration and exploitation in the search process, while the dual-indicator selection strategy is developed to enhance the population diversity. The performance of the proposed algorithm is evaluated on the DTLZ and WFG series test problems. Experimental results show that the proposed algorithm performs competitively in comparison with several MOEA/D variants. [ABSTRACT FROM AUTHOR]

Fog computing has risen into a popular topic in recent years with the idea of deploying computation and communication closer to end-users. The capability of the fog model to face serious challenges like high latency and power consumption is of paramount research significance, notably in the context of the Internet of Things (IoT), given the volume and requirements of IoT applications. Accordingly, issues such as using the capacity of the fog layer maximally, minimizing latency while maintaining reliability, and efficiently distributing the workload across the network can be well explored. This motivated us to propose a scheme for partial computation offloading and resource allocation in the fog-based IoT network with the goal of optimizing latency and power consumption (ROGI). In this research, the nodes in all network layers are involved in processing the workload. Also, the power consumption in the end-users layer is reduced. Additionally, the fog nodes collaborate to improve system reliability and provide more resources for handling users' requests. Furthermore, the concept of partial offloading is adopted, which would potentially lead to higher flexibility in resource management and provide the opportunity to leverage more resources in each layer of IoT architecture. Moreover, the whole model is decomposed into three problems, each of which is solved via optimization techniques. Extensive simulations are carried out to show the performance of the proposed scheme. [ABSTRACT FROM AUTHOR]

Cloud computing provides different types of resources to users on-demand which are hosted in cloud data centers. Aforesaid services are provided at the expense of large energy consumption. Energy consumption increases the expenditure budget, greenhouse gases, and CO2 emissions. To handle this issue, researchers have come up with various server-level energy-efficient techniques. Though the proposed techniques attempt to reduce energy consumption, they only consider the energy consumption of the CPU during the task placement process. However, researchers have recently noted that memory is also one of the higher energy consumption components and it should be considered in task placement. Moreover, existing techniques ignore the SLA violations that are encountered due to workload. To address the aforementioned issues, we propose two novel nature-inspired techniques which consider the energy consumption of both CPU and memory during the VM placement process. Proposed novel techniques are based on artificial bee colony and particle swarm optimization which haven't been used to place VM while considering energy consumption of CPU and memory. Moreover, to handle the issue of resultant SLA violations, we also provide the SLA-aware variants of the proposed energy-efficient techniques, which try to lower SLA violations faced because of excessive task consolidation. The results depict that the proposed energy-efficient techniques perform better than the existing state-of-the-art techniques, whereas proposed SLA variants also reduce the SLA violations. [ABSTRACT FROM AUTHOR]

Software-Defined Networking (SDN) is one of the promising and effective approaches to establishing network virtualization by providing a central controller to monitor network bandwidth and transmission devices. This paper studies resource allocation in SDN by mapping virtual networks on the infrastructure network. Considering mapping as a way to distribute tasks through the network, proper mapping methodologies will directly influence the efficiency of infrastructure resource management. Our proposed method is called Effective Initial Mapping in SDN (EIMSDN), and it suggests writing a module in the controller to initialize mapping by arriving at a new request if a sufficient number of resources are available. This would prevent rewriting the rules on the switches when remapping is necessary for an n-time window. We have also considered optimizing resource allocation in network virtualization with dynamic infrastructure resources management. We have done it by writing a module in OpenFlow controller to initialize mapping when there are sufficient resources. EIMSDN is compared with SDN-nR, SSPSM, and SDN-VN in criteria such as acceptance rates, cost, average switches resource utilization, and average link resource utilization. [ABSTRACT FROM AUTHOR]

Nowadays, Internet applications with different network resource requirements have been growing exponentially; therefore, the network equipment needs to be configured consistently based on application resource requirements. In this paper, Software Defined Network (SDN) is used to make the network more programmable, flexible and agile to develop the proposed model. Flow discrimination and optimized resource allocation are significant and challenging items in providing the required network resources for each flow. In this paper, we propose a model composed of (1) network flows' types discriminator and (2) the optimized resource allocation based on the flow classes. The applications are clustered into four groups according to the network resource requirements, and a deep network traffic discriminator is used for classification. The greedy algorithm is also used for optimized resource allocation. The proposed model is developed in Mininet with a Pox controller to prove Quality of Service (QoS) improvement paralleled with maximized utilization. The simulation results show that the proposed model has intelligent behaviour in network resource allocation based on flows' requirements compared to Spanning Tree Protocol (STP) and Dynamic and Adaptive Multipath Routing (DAMR) while maximizing the network utilization. [ABSTRACT FROM AUTHOR]

As a novel network architecture, network function virtualization greatly improves the flexibility of service deployment but also increases the complexity of network function virtualization resource allocation (NFV-RA). Many scholars have studied the above problems and put forward corresponding schemes. These schemes are often individually designed and are difficult to compare. A general and efficient simulation platform for the NFV-RA problem is conducive to integration, comparison and analysis of these schemes and speeds up research progress. This paper presents a general simulation platform SFCSim for NFV-RA. SFCSim adopts the discrete-time event scheduling engine, which can support the simulation of static and dynamic deployment of service function chain, service migration, mobility management and other scenarios. Finally, a load-balancing shortest path and three heuristic embedding algorithms are implemented in the platform for the service function chain embedding problem can be used as the benchmark algorithm, and SFCSim is fully compared with the Alevin to verify the performance of the platform [ABSTRACT FROM AUTHOR]

Fair and efficient allocation of multiple types of resources is a fundamental goal in cloud computing systems. Due to the heterogeneous resource requirements of a user's jobs on CPU, memory, etc., fairness of allocation is difficult to ensure while maximizing efficiency. Existing representative multi-resource fair allocation mechanisms, such as dominant resource fairness and its follow-up work, can hardly achieve maximum efficiency while ensuring fairness. To overcome this drawback, we propose a new fair allocation mechanism, called generalized asset fairness (GAF), to maximize the system resource utilization and ensure fairness. We show that GAF satisfies many desired fairness properties. To implement our resource allocation mechanism, we design a scheduling algorithm to find feasible solutions. The experimental results show that GAF can produce a fair allocation with higher resource utilization than all previous known fair mechanisms; GAF also performs well in resource sharing. [ABSTRACT FROM AUTHOR]

Efficient task processing and data storage are still among the most important challenges in Autonomous Driving (AD). In-board processing units struggle to deal with the workload of AD tasks, especially for Artificial Intelligence (AI) based applications. Cloud and Fog computing represent good opportunities to overcome the limitation of in-board processing capacities. However, communication delays and task real-time constraints are the main issues to be considered during the task mapping. Also, a fair resources allocation is a miss-explored concept in the context of AD task offloading where the mobility increases its complexity. We propose a task offloading simulation tool and approaches based on intelligent agents. Agents at the edge and the fog communicate and exchange their knowledge and history. We show results and proof-of-concept scenarios that illustrate our multi-agent-based proposition and task offloading simulation tool. We also analyze the impact of communication delays and processing units constraints on AD task offloading issues. [ABSTRACT FROM AUTHOR]

Smart resource allocation is essential for optimising cloud computing efficiency and utilisation, but it is also very challenging as traditional approaches often overprovision CPU resources, leading to financial inefficiencies. Recently developed Artificial Intelligence (AI) techniques have the potential to solve this problem efficiently; for example, deep learning models can accurately forecast how resources will be used, allowing for more efficient distribution of those resources. Despite these encouraging breakthroughs, researchers have not thoroughly investigated these AI models’ dynamic scaling potential. To address this gap, we developed a new testbed for an AI-driven cloud computing environment called CloudAIBus for effective resource allocation. CloudAIBus employs a deep learning model named DeepAR to provide a robust solution for forecasting CPU usage in order to make cost-effective resource allocation decisions. Furthermore, we implement the DeepAR model using Amazon SageMaker, a robust platform that provides the infrastructure for scalable and efficient training. We evaluated the performance of the DeepAR-based resource management approach (CloudAIBus) using Google Colab, and results show that the proposed approach offers better performance than baselines (LSTM and ARIMA-based resource management) in terms of Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Mean Squared Error (MSE). The proposed approach cut the percentage of unused CPUs from 98.65 to 32.35% compared to the GWA-T-12 dataset. This showed that it was effective at reducing over-provisioning by making accurate predictions. [ABSTRACT FROM AUTHOR]

We address the problem of resource allocation with multi-attribute demand considering heterogeneous servers in mobile edge computing (MEC). A mobile device (MD) may be within the coverage area of multiple access points and can get resources from any server that can establish direct communication. However, the configurations of heterogeneous servers are different, so the demand for each MEC server is different. We tackle this problem and consider multi-attribute demand, where each MD is assigned exactly one of multiple demands. We formulate the allocation problem in an auction-based setting to provide server scalability. We propose a family of greedy mechanisms to solve the resource allocation problem. However, each MD is self-interested and may try to manipulate the system to maximize its utility. We show our mechanisms are truthful, and they drive the system into an equilibrium where no MD can declare an untrue value to obtain higher utility. We also analyze the approximation ratios of our greedy mechanisms. Experimental results demonstrate that the greedy mechanisms obtain the near-optimal allocation in a short period while benefiting MDs and edge cloud providers. [ABSTRACT FROM AUTHOR]

Virtual machine placement (VMP) has a significant importance with respect to resource utilization in cloud data centers. Indeed, the optimized management of machine placement usually results in a significant reduction in energy consumption. VMP is a bin packing problem generalization, which is a well known hard combinatorial optimization problem. Besides being NP-hard, VMP is characterized by conflicting objectives and a noisy search space. Meta-heuristics, such as genetic algorithms, particle swarm optimization (PSO), cuckoo search (CS), tabu search and simulated annealing (SA) have been shown to be effective for this category of problems. This paper reports a performance comparison between SA, CS and PSO meta-heuristics to solve the VMP problem. In contrast to reported research work in this area, we study the performance behavior of these three meta-heuristics with respect to, not only the quality of solutions, but also the quality of the explored solution sub-space, in addition to the convergence speed towards reported solutions and the speed with which each meta-heuristic evolves towards the best reported optimized solution. Extensive simulations on randomly generated tests with sizes varying between 200 and 1000 virtual machine demands show that PSO achieves the best performance behavior with respect to all criteria. Moreover, for all tests, PSO produces a reduction of as much as 17% of the number of physical machines, 15% of the energy cost and 21% of the resource utilization of physical machines. [ABSTRACT FROM AUTHOR]

KubeEdge is an open-source system extending native containerized application orchestration and device management to hosts at the edge. One of the main disadvantages of edge computing is the lack of an effective resource allocation and privacy-preserving framework. The objective of this study dynamic task offloading is an important concept in resource allocation and privacy-preserving framework in Kubeedge-based edge computing using machine learning. This research focuses on enhancing the efficiency and security of edge computing within the framework of IoT devices within a KubeEdge cluster. It begins by gathering crucial data on computational power, memory, and network bandwidth from IoT devices, which is essential for informed decision-making. The study evaluates KubeEdge in terms of computational resource distribution and delay, introducing a privacy-preserving multi-layer Blockchain-enabled data aggregation mechanism. This approach ensures efficient data storage with adaptable and reliable data access control. The research addresses privacy concerns in IoT applications, balancing information loss and disclosure risk, and highlights the impact of forwarding traffic on cluster throughput and service delays in edge computing environments. Dynamic resource allocation algorithms are employed, considering profiling data and machine learning models for real-time task offloading decisions, guaranteeing sufficient allocated resources. A Multi-Agent Collaborative-Reinforcement Learning with Salp Swarm Algorithm is proposed for resource allocation in the edge computing environment, enhancing resource efficiency and secrecy performance. Blockchain ensures secure and transparent transactions, reinforcing trust in the dynamic edge environment. Reinforcement learning optimizes task offloading decisions, adapting to changing conditions. Auction games introduce a competitive mechanism, enhancing efficiency in resource allocation. When combined, they create a robust framework where blockchain guarantees data integrity, reinforcement learning optimizes resource usage, and auction games introduce a fair and efficient task allocation mechanism, ensuring both performance and privacy in computing environments. Furthermore, a Partitioning-Dynamic Collaborative energy-aware task offloading scheme is developed to securely offload tasks while preserving data privacy, enhancing trust computing and task offloading capabilities. The study also proposes a Hybrid Greedy Randomized Adaptive Stackelberg-Auction Game Approach to optimize offloading performance, reduce information loss, and decrease time consumption. This comprehensive framework includes task scheduling to reduce energy consumption, enhance privacy, and security, and reduce latency. The proposed work is analysed based on the Matlab software, and it shows a higher edge offloading rate and lower resource consumption for massive task scenarios in the edge network. Continuous performance evaluation of resource utilization, response time, and privacy protection further improves offloading decision accuracy over time. [ABSTRACT FROM AUTHOR]

In this paper, a design model for resource allocation is formulated beyond 5G networks for effective data allocations in each network nodes. In all networks, data is transmitted only after allocating all resources, and an unrestrained approach is established because the examination of resources is not carried out in the usual manner. However, if data transmission needs to occur, some essential resources can be added to the network. Moreover, these resources can be shared using a parallel optimization approach, as outlined in the projected model. Further the designed model is tested and verified with four case studies by using resource allocator toolbox with parallax where the resources for power and end users are limited within the ranges of 1.4% and 6%. Furthermore, in the other two case studies, which involve coefficient determination and blockage factors, the outcomes of the proposed approach fall within the marginal error constraint of approximately 31% and 87%, respectively. [ABSTRACT FROM AUTHOR]

We address the problem of online virtual machine (VM) provisioning and allocation with multiple types of resources. Formulating this problem in an auction-based setting, we propose an accurate mathematical model incorporating the ability to preempt and resume a given task for the sake of best overall use of resources. Our objective is to efficiently provide and allocate multiple VMs to maximize social welfare and encourage users to declare truthful requests. We first design an offline optimal mechanism based on the VCG mechanism; this mechanism has full knowledge of all users and offers ideal solutions. We also design an online greedy mechanism that considers only current knowledge while offering near-optimal solutions instead. Our proposed greedy mechanism consists of winner determination and payment algorithms. Furthermore, we show that the winner determination algorithm is monotonic and that the payment algorithm implements the critical payment. Both our allocation methods offer incentives to users providing true values for the sake of obtaining the best utility. We performed extensive experiments to investigate the performance of our proposed greedy mechanism compared to the optimal mechanism. Experimental results demonstrate that our proposed greedy mechanism obtains near-optimal solutions in a reasonable time. [ABSTRACT FROM AUTHOR]

In recent years, cloud computing can be considered an emerging technology that can share resources with users. Because cloud computing is on-demand, efficient use of resources such as memory, processors, bandwidth, etc., is a big challenge. Despite the advantages of cloud computing, sometimes it is not a proper choice due to its delay in responding appropriately to existing requests, which led to the need for another technology called fog computing. Fog computing reduces traffic and time lags by expanding cloud services to the network and closer to users. It can schedule resources with higher efficiency and utilize them to impact the user's experience dramatically. This paper aims to survey some studies that have been done in the field of scheduling in fog/cloud computing environments. The focus of this survey is on published studies between 2015 and 2021 in journals or conferences. We selected 71 studies in a systematic literature review (SLR) from four major scientific databases based on their relation to our paper. We classified these studies into five categories based on their traced parameters and their focus area. This classification comprises 1—performance 2—energy efficiency, 3—resource utilization, 4—performance and energy efficiency, and 5—performance and resource utilization simultaneously. 42.3% of the studies focused on performance, 9.9% on energy efficiency, 7.0% on resource utilization, 21.1% on both performance and energy efficiency, and 19.7% on both performance and resource utilization. Finally, we present challenges and open issues in the resource scheduling methods in fog/cloud computing environments. [ABSTRACT FROM AUTHOR]

Recently, modern businesses have started to transform into cloud computing platforms to deploy their workflow applications. However, scheduling workflow under resource allocation is significantly challenging due to the computational intensity of the workflow, the dependency between tasks, and the heterogeneity of cloud resources. During resource allocation, the cloud computing environment may encounter considerable problems in terms of execution time and execution cost, which may lead to disruptions in service quality given to users. Therefore, there is a necessity to reduce the makespan and the cost at the same time. Often, this is modeled as a multi-objective optimization problem. In this respect, the fundamental research issue we address in this paper is the potential trade-off between the makespan and the cost of virtual machine usage. We propose a HEFT-ACO approach, which is based on the heterogeneous earliest end time (HEFT), and the ant colony algorithm (ACO) to minimize them. Experimental simulations are performed on three types of real-world science workflows and take into account the properties of the Amazon EC2 cloud platform. The experimental results show that the proposed algorithm performs better than basic ACO, PEFT-ACO, and FR-MOS. [ABSTRACT FROM AUTHOR]

Resource provisioning is a key issue in large-scale distributed systems such as cloud computing systems. Several resource provider systems utilized preemptive resource allocation techniques to maintain a high quality of service level. When there is a lack of resources for high-priority requests, leases/jobs with higher priority can run by suspending or canceling leases/jobs with lower priority to release the required resources. The state-of-the-art preemptive resource allocation methods are classified into two classes, namely, (1) heuristic and (2) brute force. The heuristic-based methods are fast, but they can't maintain the system performance, while brute force-based methods are vice versa. In this work, we proposed a new multi-objective preemptive resource allocation policy that benefits from these two classes. We proposed a new heuristic called Best K-First-Fit (Best-KFF). The Best-KFF searches for the first k preemption choices at each physical machine (PM) and then sorts these preemption choices obtained from the PMs with respect to several objectives (e.g., resource utilization). Then, the Best-KFF selects the best choice maintaining the cloud computing system performance. Thus, the Best-KFF algorithm is a compromise between the heuristic and brute force classes. The higher the value of k is, the larger the search space is. The Best-KFF method maximizes the resource utilization of the physical machines and minimizes the average waiting time of advanced-reservation requests, the number of lease preemption, the preemption time, and energy consumption. The proposed method was thoroughly examined and compared against the state-of-the-art methods. The experimental results on various cloud computing systems demonstrated that the proposed preemption policy outperforms the state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Ultra-Reliable Low Latency Communication (URLLC) is mostly used for core services in the fifth generation (5G) and beyond cellular networks. Due to the conflicting demands of low latency and extremely high reliability, the ongoing research on this subject remains in its nascent phase. The utilization of Device-to-Device (D2D) communication shows great potential in facilitating URLLC. The implementation of URLLC and its Quality of Service (QoS) requirements may encounter high power consumption in dynamic wireless channels, potentially conflicting with the system's power limitations. In this paper, a packet delivery mechanism is introduced, aiming to achieve energy efficiency by incorporating a Power-Aware QoS-centric Preemptive Dropping mechanism to minimize latency while ensuring the successful transmission of URLLC data. The method aims to optimize bandwidth and channel assignment to effectively reduce overall power consumption. The optimization problem is formulated which is non-convex, posing challenges in selecting an optimal solution. To address this problem, we have undertaken the task of transforming the original problem into two subproblems: bandwidth allocation and subchannel assignment. To achieve optimal resource allocation for the formulated problem, we propose the Optimal Bandwidth Allocation (OBA) and Optimal Sub-channel Assignment (OSA) algorithms. The simulation results validate the proposed method outperforms the benchmark schemes, demonstrating higher energy efficiency ranging from 6.96 to 31.6%, as well as a reduction in latency ranging from 5.7 to 25.67%. The proposed method exhibits broad applicability in various domains, including remote healthcare and Industrial Internet of Things (IIoT) applications. [ABSTRACT FROM AUTHOR]

With the continuous expansion of the scale of the industrial Internet, edge computing has become an indispensable part of the industrial Internet. In order to quickly handle the massive computation tasks of production devices and monitoring devices in industrial production and ensure the safety and efficiency of industrial production. This article considers the Joint Task Offloading and Resource Allocation (JTORA) optimization problem, which is measured by the weighted sum of task completion time and energy consumption, and includes the joint optimization of task offloading decision, uplink power allocation, and computing resource allocation. Also further decompose the JTORA problem into (i) a Resource Allocation (RA) problem with fixed task offloading decision and (ii) a Task Offloading (TO) problem that optimizes the optimal-value function corresponding to the RA problem for hierarchical optimization. This paper adopts Deep Reinforcement Learning (DRL) algorithm to solve the RA problem and a heuristic algorithm for the TO problem. Simulation experimental results show that the proposed JTORA optimization scheme can significantly reduce the production time and device energy consumption in industrial production over traditional approaches. [ABSTRACT FROM AUTHOR]

A wide range of applications in distributed environments have a linear structure, varying priorities, and may experience transient software failures. As the computational demands of such linear workflow (LW) jobs continue to grow, their efficient, fair, and fault-tolerant resource allocation and scheduling is becoming more challenging. To address this problem, we propose a fair and efficient scheduling approach, which considers that the priorities of the jobs age with time. We jointly use this scheduling strategy with three practical routing techniques, as well as two variants of an application-directed checkpointing scheme. The first variant of this scheme incorporates imprecise computations in a selective manner, whereas the second one does not use imprecise computations at all. Our aim is to dynamically allocate and schedule LW jobs with different priorities and transient software failures in a distributed system. Through extensive experimentation, we evaluate the system performance under the considered routing methods and checkpointing schemes, utilizing various load cases and failure probabilities. The simulation results showcase the impact of selective imprecise computations on the system performance, while providing insights into how the examined routing strategies perform in each of the investigated scenarios. [ABSTRACT FROM AUTHOR]

In the past decade, the cloud environment has emerged as a resource-enriched technology for computational modeling. Most cloud data centers are deployed far from the device layer, which may cause higher service delivery time. Therefore, Fog servers are implemented to retain the commitment of minimum service delivery time. The computational resources (processing unit, memory, storage) and network bandwidth are to be appropriately scheduled to make an efficient system. For delay and resource-constrained devices, the essential issues include minimum service latency and improved system performance by offloading non-sensitive devices to the cloud layer. In this study, an adaptive bandwidth adjustment technique is proposed considering the sensitivity of the applications and load on the multiple queues at the access point based on which, a compelling dynamic bandwidth ratio is calculated. Further, blocking probability of task and task dropping probability is used as task offloading probability to overcome the dropping ratio of type-II applications. The random early detection (RED) algorithm is implemented in queueing management to maximize the system performance. Extensive simulation results highlighted that the proposed model significantly improves the system performance in terms of network delay, queue delay, and overall response time over traditional algorithms. [ABSTRACT FROM AUTHOR]

Time-varying resource allocation allows users to define their own unique resource requirement plans during different time periods. This mode of allocation can increase the flexibility of resource usage and reduce resource usage costs for users. Moreover, combining this approach with an auction mechanism can enable resource providers to obtain greater social welfare and benefits; therefore, such resource allocation has become a hot topic in cloud computing. This paper addresses the problem of time-varying batch virtual machine (VM) allocation and pricing in the cloud. Specifically, (1) we propose a novel integer programming model for the time-varying batch VM allocation problem, and (2) we design two truthful auction mechanisms to solve the allocation and pricing problem in a competitive environment. The optimal mechanism includes a dynamic programming (DP)-based resource allocation algorithm and a Vickrey–Clarke–Groves (VCG)-based payment price algorithm. Meanwhile, we also design a greedy mechanism that includes a dominant-resource-based allocation algorithm and a dichotomy-based payment price algorithm. We prove the economic characteristics, including truthfulness and individual rationality, of the above two mechanisms. Furthermore, we prove the approximation ratio of the allocation algorithm in the greedy mechanism. Compared to state-of-the-art research, our approach is characterized by high social welfare, a high served user ratio and a short execution time. [ABSTRACT FROM AUTHOR]

Cloud computing is a new technology, providing different online resources and services to users. Load balancing has become an interesting research area in this field. A few existing scheduling algorithms can maintain load balancing and provide better strategies through efficient job scheduling and resource allocation techniques as well. In order to gain maximum profits with optimized load balancing algorithms, it is necessary to guarantee a service level along dimensions such as performance, availability and reliability. In this work, we propose a Model-based approach to study and analyze centralized and distributed dynamic load balancing protocols in the Cloud. The formal verification of different properties of the studied protocols has been performed automatically and a performance analysis allowing their comparison is also provided. In this paper, we show how aspects of performance, resource consumption, and reliability of the Cloud can be formally modeled, verified and analyzed using a component-based architecture. [ABSTRACT FROM AUTHOR]

Elasticity is an important feature of cloud computing, which allocates/de-allocates adequate computing resources automatically and provisions and de-provisions computing resources timely when the workload fluctuates. It can help in better resource utilization. Elasticity can address the challenges of limited physical resources such as cost, computing power, storage capacity, etc. In cloud computing, the cloud providers achieve elasticity through virtualization techniques such as virtual machines (VMs) or containers. Elasticity can help to scale-up/down the various computing resources of a running system when the workload increases or decreases over the machine. In this paper, we propose a hybrid model that uses a container-based virtualization techniques allocated for various computing resources of a running system (container). It uses a multi-criteria decision making procedure for the allocation of resources. This model uses a reactive approach for scaling up and scaling down the containers' resources as per the requirements of the application. The proposed approach provides better resource utilization and minimizes subscription costs from both the provider and consumer perspectives. The experimental results show that the hybrid model dynamically allocates resources to applications instantly and maintains higher resource utilization than horizontal elasticity, vertical elasticity, and Docker. [ABSTRACT FROM AUTHOR]

This research paper proposes a novel approach named priority-based load balancing (PLB) for cloud computing environment. The PLB provides a resilient and adaptive task scheduling using multi-queues. Numerous strategies have already been proposed in the past researches to prioritize the tasks and mapping all the tasks to different resources available on the cloud. There is still a hindrance in the performance due to the negligible attention paid to the unused resources and tasks having low priority, eventually leading to starvation problem. To this end, the PLB algorithm has been partitioned into four sub-procedures, namely (i) Starvation-free task allocation, (ii) Inserting tasks into the dispatcher, (iii) Re-ordering tasks inside the queues and eventually, (iv) Mapping tasks onto the Virtual Machines (VMs) calculating the cost incurred for all the corresponding VMs. The sole motivation of this research work is to optimize the performance parameters by allocating all the jobs to all the available resources in the workflow model. It also consolidates the job categorization in the priority-based multi-queues, while filtering tasks from all the queues to overcome the deprivation of low priority tasks. In this paper, a test-bed setup has been deployed using CloudSim 3 and TCS WAN emulator for experimentation and results evaluation. The experimental setup imbibes different aspects such as performance measures, average response time, makespan time in order to ascertain efficiency, resource utilization ratio and bandwidth of the workflow model. The obtained results are further compared with five different approaches including- First Come First Serve, Round Robin, Min–Min, Max–Min and ACO and it was observed that the proposed strategy yielded more efficiency and accuracy in most of the cases. The experimental results have been further validated and demonstrated in order to justify the claims of the proposed approach, being able to tackle out different priority tasks and resource allocation in a stable and optimum manner. [ABSTRACT FROM AUTHOR]

Infrastructure as a Service (IaaS) is a cloud computing service provided over the internet to facilitate the provisioning of various services such as storage, processes, etc. The provider in the IaaS market may offer some purchasing plans including: reservation, on-demand, and spot plans for its resources. As in real scenarios, demand volume for each plan is assumed to be a random variable with a given probability distribution. The provider maximizes its average revenue in the long run by optimal allocation of its resources among the plans. We formulate an Integer Linear Programming (ILP) model with a stochastic constraint, to determine the number of resources to be allocated for each plan in every time slot in the planning horizon. First, fixed prices are considered for each plan, then two mechanisms of Continuous Double Auction and Second Price Sealed Bid Auction are considered for reservations and spot plans, respectively, to obtain market-driven prices of the services. The Seasonal Weighted Moving Average method is used to predict the amount of demand in every slot. Finally, the proposed mechanisms are evaluated through simulations and the results confirm the effectiveness of the methods in maximizing the revenue and overall utilization of the available IaaS capacity. [ABSTRACT FROM AUTHOR]

With the penetration of Network Function Virtualization (NFV), network functions, traditionally deployed as proprietary physical equipment like firewalls, Network Address Translations (NATs), are gradually being implemented as software and deployed on standardized hardware. One of the crucial challenges in this paradigm is how to place the software implemented network functions to minimize the number of used physical servers. In this paper, we study the problem of how to optimally place Virtual Network Functions (VNFs) in networks where it is allowed to migrate already placed VNFs to decrease used servers. We first formulate the offline problem as an Integer Linear Programming (ILP) problem, and then propose a semi-online algorithm to solve the online variant of the problem. We name the proposed algorithm Semi-onlIne Vnf plAcement (SIVA). In particular, SIVA is based on a bin packing algorithm that solves online bin packing problem while taking care of migrations. According to our theoretical analysis, SIVA migrates at most λ VNFs each step, and it has Asymptotic Competitive Ratio (ACR) of 3/2 if k → ∞ , where λ = k · | N | , k is a tunable parameter, and | N | is the the number of supported VNF types. We conduct extensive numerical simulations to evaluate the performances of SIVA. The experiment results validate the theoretical analysis and show that SIVA outperforms the state-of-the-art algorithms by achieving near-optimal performance with minor VNF migrations. [ABSTRACT FROM AUTHOR]

Cloud computing has attracted scientists to deploy scientific applications by offering services such as Infrastructure-as-a-service (IaaS), Software-as-a-service (SaaS), and Platform-as-a-Service (PaaS). The research community is able to get access to resources on-demand within a short period of time. But, as the demand for cloud resources is dynamic in nature, this affects resource availability during scheduling. Hence, there is a need for efficient management of resources so that tasks can be scheduled based on their execution requirements. To provide a solution, a resource prediction based scheduling approach has been introduced in this paper which automates the resource allocation for scientific applications in a virtualized cloud environment. This research work focuses on the design of an optimized prediction based scheduling approach which maps the tasks of scientific application with the optimal VM by combining the features of swarm intelligence and TOPSIS. The proposed approach minimizes the execution time, cost, and SLA violation rate in comparison to existing scheduling heuristics. [ABSTRACT FROM AUTHOR]

The strong computing power of Cloud computing enables rapid processing of mass data. Hadoop, the most extensively-distributed architecture among Cloud-computing platforms, uses the MapReduce programming model in a server cluster to separate the applications into many small parts in order to conduct arithmetic processing of big data. Virtualization technology renders the various computer physical resources into abstract virtual resources; thus, computer resources are used more flexibly to simplify management. In practice, Hadoop is frequently combined with virtualization technology to reduce handling costs; however, the virtualized Resource Pooling characteristic results in hardware access conflicts and performance degradation. This study revealed where virtualization degrades read and write capabilities by comparison of the substantialistic architecture and virtualized architecture, and strategies for solving conflicts in virtualization are presented. An optimized virtualized Hadoop architecture was strategically designed; conflicts were gradually solved, and the performance of the virtualized Hadoop was enhanced. The experimental results proved that the strategies presented solved the conflicts while maintaining the fault tolerance, processing performance, and expandability of Hadoop. [ABSTRACT FROM AUTHOR]

The dynamic resource allocation is a good feature of the cloud computing environment. However, it faces serious problems in terms of service quality, fault tolerance, and energy consumption. It was necessary, then, to find an effective method that can effectively address these important issues and increase cloud performance. This paper presents a dynamic resource allocation model that can meet customer demand for resources with improved and faster responsiveness. It also proposes a multi-objective search algorithm called Spacing Multi-Objective Antlion algorithm (S-MOAL) to minimize both the makespan and the cost of using virtual machines. In addition, its impact on fault tolerance and energy consumption was studied. The simulation revealed that our method performed better than the PBACO, DCLCA, DSOS and MOGA algorithms, especially in terms of makespan. [ABSTRACT FROM AUTHOR]

Attracting customers through reward programs is the primary key success for customer-oriented organizations. One of the most famous customer reward programs is applying price discount. In our negotiation-based cloud resource allocation problem, discount price is offered based on both status of a provider and behavior (or loyalty class) of a customer. That is, a resource customer who has appropriate buying behavior and negotiates for resource type instances with high necessity to sell is deserved to receive high price discount from provider. To do this, first, three customer's loyalty classes are defined and customers are classified into theses classes in terms of their previous buying behavior using fuzzy system in name FCLCDS. Second, another fuzzy system in name FNTSVMTDS is designed to determine the value of necessity to sell resource type. The outputs of FCLCDS and FNTSVMTDS are called Loyalty Class (LC) and Necessity to Sell VM Type (NtSVMT), respectively. Finally, a fuzzy system in name FDCDS is proposed to determine the discount coefficient based on both LC and NtSVMT inputs. Furthermore, appropriate times for calculating/re-calculating the discount coefficients that are applied by a resource provider to relax its counter-offers are calculated. We perform extensive simulation experiments to compare our designed negotiator in name FDMDA with MDA and FNSSA. The results show that our designed FDMDA outperforms MDA and FNSSA. [ABSTRACT FROM AUTHOR]

Cloud computing is emerging as an increasingly popular computing paradigm, allowing dynamic scaling of resources available to users as needed. This requires a highly accurate demand prediction and a resource allocation methodology. The existing methodologies for dynamic resource allocation do not provide effective performance isolation between the VM and Artificial Demand Analysis machines since it gets affected by interferences. To overcome these issues, this paper proposes a conceptual model and an effective algorithm to achieve dynamic resource allocation by migrating tasks or requests in VMs. At first, task demands from the multiple users go to the feature extraction process. In feature extraction, features of the user's tasks and cloud server are extracted. Next both features are reduced by using Modified PCA algorithm to reduce the dynamic resource allocation processing time. Finally, both the features are combined and resource allocation is performed using Hybrid Particle Swarm Optimization and Modified Genetic Algorithm (HPSO-MGA). Then the optimized task has been scheduled to particular VM for allocating the resources. The experimental result of the proposed resource allocation methodology indicates better performance when compared with the existing methods Firefly and Krill herd Load Balancing (LB). For 100 VMs the reliability of HPSO-MGA is 0.87 but the exiting krill herd LB and IDSA gives 0.78 and 0.85, which is lower than the proposed one. [ABSTRACT FROM AUTHOR]

Serverless computing and a function execution model, Function-as-a-Service (FaaS), are currently receiving considerable attention from both academia and industry. One of the reasons for the success of serverless computing is its straightforward interface that abstracts complex internals of cloud computing resource usage and configurations. However, this approach may result in hiding too much information about how underlying cloud resources would work, entailing that users cannot predict how their applications will perform, especially for IO-heavy ones. To address this issue, we evaluate several aspects of network and disk IO performance with realistic workloads using public FaaS systems. Our analysis reveals that current public FaaS systems do not provide appropriate levels of IO performance differentiation, and the ability to isolate network resource allocation during concurrent execution is rarely offered by service providers. Based on the results presented in this paper, we insist that it must be mandatory for network and disk IO resource performance of FaaS to be more visible and predictable, as is the case for memory and CPU, in order to expand serverless computing applications to data-intensive ones. [ABSTRACT FROM AUTHOR]

Software-defined networks (SDNs) are a huge evolution in simplifying implementation and network operation which have reduced costs and made the network programmable. Although SDNs are a suitable option for solving some of the previous problems, but they have some challenges as any other new technology. Resource allocation and balance control in network is one of the main challenges of this technology which is studied in this paper. In this study, a new approach is proposed for improving memory resource allocation in network using load distribution clusters. Since in the proposed method, K-mean++ algorithm is used for clustering, load balancing of clusters can be used to preserve load balance of the network. In the proposed method, data with higher recall is transmitted to high-quality clusters in terms of average number of hubs and lower average delay between server and user. In the proposed method, by increasing number of clusters, higher memory is created in the network. [ABSTRACT FROM AUTHOR]

Dynamic resource allocation is the key objective of the paper motivated due to a large number of user's service request and increasing network infrastructure complexity. Load balancing and Service Broker Policy are taken as two main key areas for the dynamic provision of resources to the cloud user in order to meet the QoS requirement. While provisioning the resources, the conventional approaches degrade due to QoS performance limits such as time delay, energy, etc. To overcome those problems, we proposed a new approach to provide dynamic provisioning of resources based on load balancing and service brokering. Initially, the Multi-agent Deep Reinforcement Learning-Dynamic Resource Allocation (MADRL-DRA) is used in the Local User Agent (LUA) to predict the environmental activities of user task and allocate the task to the Virtual Machine (VM) based on priority. Next, a Load balancing (LB) is performed in the VM, which increases the throughput and reduces the response time in the resource allocation task. Secondly, the Dynamic Optimal Load-Aware Service Broker (DOLASB) is used in the Global User Agent (GUA) for scheduling the task and provide the services to the users based on the available cloud brokers (CBs). In the global agent, cloud brokers are the mediators between users and providers. The optimization problem in Global Agent (GA) is formulated by the programming of mixed integers, and Bender decomposition algorithm. The result of our proposed method is better as compared with the conventional techniques in terms of Execution Time, Waiting Time, Energy Efficiency, Throughput, Resource Usage, and Makespan. [ABSTRACT FROM AUTHOR]

Delivering relief supplies to victims of natural disasters is a very complicated process faced with many challenges: damaged road network, high demand for various materials, scarcity of resources, multi-supply and multi-demand sites, limited transport capacity, etc. Traditionally, outreach maximization is the focus of emergency relief. In this research, we propose a bi-level programming model which takes the aforementioned challenges into full account. We consider two objectives: minimizing the distribution time, and maximizing the allocation fairness. Because the problem is NP-hard, we design an improved differential evolution (IDE) algorithm and numerically compare it with several conventional differential evolution algorithms, including CoDE, SaDE, JADE and JDE. We found that the proposed IDE outperforms the existing algorithms. The feasibility and validity of the proposed model and the IDE algorithm are verified by applying them to the 2008 Wenchuan earthquake in western China. [ABSTRACT FROM AUTHOR]