查看更多>>摘要：The 'replica placement and migration' mechanism for software-as-a-service (SaaS) developments in the edge cloud is developed. The placement of replica problem is rectified by utilising the hybrid position of wild geese and golden tortoise beetle (HPWGTB). For the similar data module, the different replicas should be placed on different data nodes. The multi-objective constraints such as network transmission cost, node load, and file unavailability are considered for an effective replica placement and migration. The developed hybrid HPWGTB is utilised to improve the load balancing of data nodes, decrease the response time, and reduce the resource utilisation of networks. The migration relationship between the target node and source node is considered for developing a migration of replica approach for accessing hotspots and minimising the migration time. The experimental outcomes are validated by comparing them with other optimisation approaches.
查看更多>>摘要：Task scheduling in cloud computing environment becomes difficult when complexity level of dispute, including task count and computing resources, rises with user growth. Solving this, an enhanced task scheduling (ETS) model with optimised artificial intelligence driven by swarm is proposed in paper. In proposed method, supervised machine learning algorithm, artificial neural networks (ANN) with swarm-based moth flame optimisation (MFO) is used to balance scheduling. MFO optimises by separating out virtual machines (VMs) considering basic properties like CPU utilisation, memory and bandwidth. ETS model is optimised based on resource allocation and balancing issues using back-propagation algorithm (BPA) with ANN (ANN-BPA) to analyse scheduling and problem identification mechanism. Efficiency of ETS model is assessed, focusing on aspects such as task allocation, task completion, execution time and energy consumption. The ANN-BPA-based task scheduling model outperforms by present technique and ANN-based model, which enhances resource utilisation by 7.54% and decreases completion time by 0.6 s.
查看更多>>摘要：The need for smart homes with many devices and services continues to rise quickly. With this surge, smart homes need task scheduling and load-prediction algorithms to provide the proper services for the residents. A deep learning-based dynamic job scheduling and load prediction technique for cloud-fog smart homes is proposed in this paper. This algorithm forecasts task arrival rates at each fog node and assigns them to available fog nodes. It dynamically schedules tasks based on fog node workload. Another option is to send non-real-time jobs to the cloud and real-time tasks to the fog layer. This optimises load distribution for performance. Using these task assignee models and features, the program optimises prioritised tasks, scores, network latency, and device resource characteristics. We simulate the algorithm's performance in various workloads in this part. The proposed algorithms achieved in higher percentile for 93.79% latency, 95.00% throughput, 95.34% response time, 96.28% scalability, 94.20% fault-tolerance, 97.41% scheduling capacity, 91.41% load balancing capacity, and 95.22% priority management. The results indicate that such an algorithm significantly surpasses the conventional task scheduling methods in load balancing and shortens the average task response time.
查看更多>>摘要：A secure framework for storing EHR in the hybrid cloud platform is implemented. At first, the required medical data is gathered from the database of the hospitals and split into sensitive and insensitive parts. The sensitive part is first encrypted with the data sanitisation method. The keys obtained are optimally chosen by the modified uniform number-based red fox optimisation (MUN-RFO). Then, the insensitive part is encrypted by optimal key-based data encryption scheme (OKDES), in which the same MUN-RFO algorithm is utilised to choose the optimal keys. The sensitive and insensitive data is combined and stored in the Hybrid Cloud. From the result analysis, the cost function of the MUN-RFO-OKDES is lower by 0.022% of HBA-OKDES, 0.011% of DHOA-OKDES, 0.0025% of EFO-OKDES and 0.001% of RFO-OKDES at 5th iteration for dataset 2. Numerous simulations are carried out to prove the security provided by the given data storage model in a hybrid cloud.
查看更多>>摘要：The stateless nature of serverless computing makes it a viable choice for establishing the long-desired edge-cloud continuum. Current efforts to provide a unified view over both the cloud and the edge are vendor-centric, with proprietary interfaces. This makes interoperability between different infrastructures difficult, while also raising questions about future-proofing. To overcome this problem, we introduce a federation layer which provides a unified view over distinct edge-cloud solutions. We summarise the current open questions and define a set of functional requirements for a unifying federation layer. We identify the main pillars required for fulfilling the requirements from a technical perspective and discuss concrete implementation approaches. Finally, we also showcase a practical verification of the architecture, leveraging a federation of geographically distributed Kubernetes clusters. The verification is done using multiple serverless runtime options, with computing environments scattered both in the cloud and in the edge, overcoming various connectivity restrictions in place.
NETL
NSTL