No abstract

Network virtualization is an inherent component of future internet architectures. Network resources are virtualized from the underlying substrate and elastically provisioned and offered to customers on-demand. Optimal allocation of network resources in terms of utilization, quality of service, and energy consumption has been a challenge. Existing solutions consider congestion control in a single-objective virtual network embedding (VNE) problem. This paper defines a multiple-objective VNE problem called the congestion-aware, energy-aware VNE (CEVNE). The aim is to seek a solution that saves cost, saves energy and avoids network congestion simultaneously. CEVNE's modelling techniques and solution approaches apply both the weighting method and the constraint method to search for paretooptimal solutions that produce the best compromised solutions for all three objectives. Solving VNE problem is, however, NP-hard. A heuristic solution is proposed involving a two-stage coordinated CEVNE. The node-mapping algorithm searches for the suboptimal solutions for three objectives. The link mapping process is an SDN-based heuristic algorithm that deploys a path service and a resource monitoring application on an SDN controller. The solution is realized using SDN, Segment Routing, and open network operating system platform (ONOS) technologies. The energy minimization is implemented with a registry that keeps track of active nodes and sets inactive nodes to sleep mode. The evaluation results showed that the multiple-objective CEVNE approach is feasible and achieves its goals of optimizing the resource allocation, improving the runtime, saving the energy consumption and controlling the network congestion.

No abstract

The paper presents a comprehensive overview of intelligent video analytics and human action recognition methods. The article provides an overview of the current state of knowledge in the field of human activity recognition, including various techniques such as pose-based, tracking-based, spatio-temporal, and deep learning-based approaches, including visual transformers. We also discuss the challenges and limitations of these techniques and the potential of modern edge AI architectures to enable real-time human action recognition in resource-constrained environments.

Many health professionals do not use correct person transfer techniques in their daily practice. This results in damage to the paraspinal musculature over time, resulting in lower back pain and injuries. In this work, we propose an approach for the accurate multimodal measurement of people lifting and related motion patterns for ergonomic education regarding the application of correct patient transfer techniques. Several examples of person lifting were recorded and processed through accurate instrumentation and the well-defined measurements of kinematics, kinetics, surface electromyography of muscles as well as multicamera video. This resulted in a complete measurement protocol and unique reference datasets of correct and incorrect lifting schemes for caregivers and patients. This understanding of multimodal motion patterns provides insights for further independent investigations.