Modelling and simulation of cloud‐native‐based edge computing terminals for power distribution

Docker as the representative container technology promotes the rapid development of container cloud with its flexible, efficient and fast characteristics. As a better resource virtualization technology, container is widely recognized and favored by the industry. The resource scheduling strategy in the container cloud has an important impact on the cluster performance. The default priorities strategy of Kubernetes container orchestration tool only considers CPU and memory. With the complexity of the production environment, only considering the above two factors in the priorities stage can easily lead to some node load imbalance, which is difficult to guarantee the cluster service performance. In order to balance the resources of container clusters, this paper proposes a new priorities stage strategy based on Kubernetes' default scheduling strategy, which takes the storage and network bandwidth of task nodes into account, then calculates the weight of each resource, and finally puts it into the scoring formula as the basis of resource scheduling. By comparing the load balancing rate of the default priorities scheduling algorithm and the improved priorities scheduling algorithm, it is verified that the improved container cloud resource scheduling strategy can make the load of the task nodes in the cluster more balanced.

As machine learning (ML) models are used more and more for important business tasks, it is important to make sure that ML operations are safe. This makes it easier to automate tasks, make decisions, and come up with fresh ideas in a number of fields. These procedures generally comprise sensitive datasets, proprietary algorithms, and complex pipelines that go through many phases, such as taking in data, preparing it, training it, validating it, and deploying it. These pipelines are open to data corruption, model changes, and illegal access since they don't have strong security measures in place. All of these things might make the models substantially less accurate and reliable. Kubernetes is the best platform for building and scaling machine learning operations because it has strong container orchestration features, declarative configuration, and works with the latest MLOps tools. It is always changing and falling apart, however, which means it has its own security problems that need to be dealt with carefully and with careful preparation. This article talks about how to use Kubernetes and follow the best practices and principles of the Certified Kubernetes Security Specialist (CKS) framework. This post's purpose is to provide you knowledge on how to make machine learning pipelines safer. We will talk about how to set up strong Role-Based Access Control (RBAC) based on the principle of least privilege, how to create detailed network policies to keep workloads separate, how to safely manage secrets using both native Kubernetes tools and third-party tools, and how to use runtime security measures like container image scanning and admission controllers to make sure that everything stays safe at all times. In the context of building machine learning systems that can solve problems, we want to stress how important DevSecOps concepts are. This list includes things like putting security first, automating the process of finding vulnerabilities, and always keeping an eye on pipelines. In this case, the security rules that come with Kubernetes are applied. Companies may build machine learning systems that are scalable, auditable, and safe utilizing methods that CKS has approved. This lowers risks and encourages new ideas. There is a case study at the end of the post that shows how a tiered security policy may make a real-world machine learning pipeline better. This is solid advice for anybody who is working with Kubernetes-based machine learning systems when it comes to putting security first

Challenge of PWR new core design simulation: A focus on uncertainties due to nuclear data and reflector modelling

Performing an in-depth study on the Power Internet of Things (IoT), it is known that the distribution network faces higher security risks due to a large number of terminals, various types of equipment, and the flexible access of each device to the distribution network structure and dynamic boundary. Therefore, it is urgent to make significant research on the security control mechanism for the power distribution Internet of Things, to realize the safe access of IoT devices in the Power IoT. Firstly, this paper analyses the shortage of power system security under the background of the IoT. Secondly, trusted computing technology is represented by the TrustZone and is deeply studied, and the feasibility of its application for the security of power distribution IoT is analyzed. Furthermore, the TrustZone-based security management and control architecture for power distribution novel IoT equipment is proposed, and its integrity measurement and trusted network connection process are analyzed. Finally, the security of this architecture is analyzed briefly.

BackgroundNew mathematical models of complex biological structures and computer simulation software allow modelers to simulate and analyze biochemical systems in silico and form mathematical predictions. Due to this potential predictive ability, the use of these models and software has the possibility to compliment laboratory investigations and help refine, or even develop, new hypotheses. However, the existing mathematical modeling techniques and simulation tools are often difficult to use by laboratory biologists without training in high-level mathematics, limiting their use to trained modelers.ResultsWe have developed a Boolean network-based simulation and analysis software tool, ChemChains, which combines the advantages of the parameter-free nature of logical models while providing the ability for users to interact with their models in a continuous manner, similar to the way laboratory biologists interact with laboratory data. ChemChains allows users to simulate models in an automatic fashion under tens of thousands of different external environments, as well as perform various mutational studies.ConclusionChemChains combines the advantages of logical and continuous modeling and provides a way for laboratory biologists to perform in silico experiments on mathematical models easily, a necessary component of laboratory research in the systems biology era.

Container orchestration has emerged as the foundation of modern cloud-native application development, fundamentally transforming how distributed systems are designed, deployed, and managed across enterprise environments. Kubernetes, an open-source container orchestration platform, has emerged as the de facto standard for scaling containerized applications and sits at the center of this transformation. The platform's core components, including Pods, Services, and Deployments, establish the foundational architecture for microservices communication and lifecycle management. Advanced deployment strategies encompassing rolling updates, blue-green transitions, and canary releases enable organizations to maintain continuous service availability while implementing frequent code changes. The horizontal and vertical autoscaling capabilities automatically adjust resource allocation based on real-time demand patterns, ensuring optimal performance during traffic fluctuations. Cluster autoscaling complements application-level scaling through intelligent infrastructure provisioning and cost optimization. Comprehensive monitoring and configuration management features provide deep operational visibility while maintaining security and compliance standards. The declarative configuration model eliminates manual intervention requirements, reducing operational overhead while improving system reliability. Health checking mechanisms through liveness and readiness probes ensure automatic failure detection and recovery. The platform's extensive ecosystem integration capabilities support diverse monitoring, logging, and tracing solutions, creating comprehensive observability across distributed application infrastructures and enabling proactive performance optimization strategies.

Bus bars are very critical components for power distribution in Electric Drive systems. The increase of power electronic components on vehicles, and their interaction, has increased the possibility of noticeable and destructive amplitude harmonics at resonant frequencies to exist in the power distribution network. The presence of stray inductance of bus bars along with the capacitance of the inverters creates a resonant circuit, which in turn can create instability in the system, so it is very critical to understand the performance and placement of Busbars in the system. As impedance is very much influenced by the geometry, shape, and materials of the bus bars, simulation tools can be utilized to accurately predict the impedance and this can be utilized to study the performance of the complete system and further optimize the busbar geometry as per system needs. This paper will discuss the virtual modeling procedure established using FEM for calculating inductance based on the magnetic field and current distribution obtained from the magnetic field analysis and further utilizes this data for system optimization.

A magnetic induction based underwater communication link is evaluated using an analytical model and a custom Finite-Difference Time-Domain (FDTD) simulation tool. The analytical model is based on the Sommerfeld integral, and a full-wave simulation tool evaluates Maxwell's equations using the FDTD method in cylindrical coordinates. The analytical model and FDTD simulation tool are then compared and used to predict the system performance for various transmitter depths and optimum frequencies of operation. To this end, the system bandwidth, signal to noise ratio, and the magnitude of the induced voltage are used to estimate the expected channel capacity. The models show that in seawater, a relatively low-power and small coil may be capable of obtaining a throughput of 40 to 300 kbps, for the case where a transmitter is at depths of 1 to 3 m and a receiver is at a height of 1 m.

The complexity of networks today makes it difficult to handle fault restoration by means of human intervention. Future network architectures are expected to be self-protecting and, more generally, self-organizing. In this article we describe the modeling methods and simulation tools we have used for the analysis of a new integrated restoration scheme operating at multiple layers/networks. The networks over which we have performed the analysis are ATM and SDH, but the methodology could be applied to other multilayer architectures too. A network model has been created that consists of both ATM and SDH layers. Using this network model, we have seen how the new scheme explores the free bandwidth over both layers. The multilayer bandwidth exploration was modeled using Microsoft Visual C++. Using OPNET Modeler, a node model was created to analyze the restoration message processing within it and the interaction with other nodes for the purpose of simulating the restoration delay. The article emphasizes the simulation methodology rather than the results of the simulation. Among others, we describe which additional functionalities a simulation tool is necessary to have in order to simulate multilayer networks. A comparison of OPNET and Microsoft Visual C++ is made, describing their advantages and the reasons we found it necessary to use both.

Modern particle accelerator facilities require sophisticated control systems capable of managing thousands of process variables in real-time while ensuring high availability, scalability, and ease of maintenance. This paper presents EPIK8S (EPICS on Kubernetes), a framework that leverages Kubernetes container orchestration to deploy, manage, and scale EPICS (Experimental Physics and Industrial Control System) infrastructure for accelerator control systems. A central design principle is that the entire control system configuration for a beamline — comprising IOCs, services, and infrastructure — is captured in a single YAML file, which is processed through a Jinja2 templating layer to produce device-specific IBEK runtime configurations. This three-tier architecture radically simplifies IOC management, enables complete change tracking via Git, and eliminates an entire class of manual configuration errors. The framework introduces a GitOps-based approach using ArgoCD for continuous deployment, providing declarative configuration management and automated synchronisation. We describe the architecture and implementation, and report operational experience from three INFN facilities: the SPARC_LAB photoinjector, the DAFNE Beam Test Facility (BTF), and the ELI-NP gamma beam system in Romania. Performance metrics — including dedicated Channel Access round-trip latency measurements comparing bare-metal, pod-to-pod, and external-to-cluster scenarios — and lessons learned from over two years of production operation demonstrate significant improvements in deployment efficiency, maintainability, and reliability compared to traditional bare-metal EPICS deployments, with containerisation overhead well within acceptable bounds for accelerator control applications.

Deepwater corals (Lophelia) was identified in the vicinity of the location of the exploration well Pumbaa on the Norwegian Continental Shelf where a drilling operation was carried out by the GDF SUEZ E&P Norge AS (GDF SUEZ). When drilling in sensitive areas, it is a requirement from the Norwegian authorities that the operator shall monitor the drilling operation. Possible impacts on the corals were simulated prior to the drilling operation. In addition, monitoring of the discharges during drilling was carried out in order to document effects on the corals. The paper considers a comparison between measured and simulated discharges of drill cuttings and mud to the sea on the Pumbaa field in the Norwegian Sea. The actual monitoring program is presented in a separate paper on this conference (Moskeland et al. 2012). The main objective of the monitoring was to assess effects from the drilling activities on the corals south (and nearest) of the drilling location. The monitoring program included measurements of current, turbidity and sedimentation, as well as sampling of sediments from the seafloor. The data collected in the monitoring program turned out to be useful as well to validate numerical models for simulations of the discharges. In the present case, corals are located close to the drilling site, and the simulation tool could then be validated for prediction of potential exposure of the corals. The actual comparison between model simulations and measurements is carried out by comparing the actual depositions of barite on the sea floor. In addition, turbidity recordings close to the sea floor are compared with simulated concentrations of particle matter at the locations of the turbidity recordings. Good correspondence is obtained between model results and measurements on the occurrence of events of particle matter present in the water column at selected locations. A good similarity is found between measured/simulated depositions of barite at the sea floor, and measured/simulated concentrations of particle matter in the water column. Use of the model validated will strengthen the operator's ability to predict exposure (and consequently potential damage) on corals (and other biological resources) prior to the drilling operation. The operator may then obtain an "early warning" on exposure (potential damage). Measures can then be initiated to eliminate or reduce environmental impact.

Heat pipe failure accident analysis in megawatt heat pipe cooled reactor

Public wireless network has been widely utilized in power distribution communication due to its benefits of ease deployment and flexible coverage. This paper proposes a reliability enhancement scheme for remote control services in power distribution, and verifies its effectiveness through experiments in practical power distribution scenarios. The scheme utilizes double network based dual-mode and secure encryption devices to achieve dual network availability. With the secure access zone connected to the distribution station and the public wireless network equipment management system deployed, the online ratio of the power distribution terminals would increase by introducing the redundancy in the operation of public wireless communications. Experiments are conducted to remotely control FTUs in practical power lines, and the results demonstrate expected performance.

Containers simplify the packaging, deployment and orchestration of diverse workloads on distributed infrastructure. Containers are primarily used for web applications, databases, application servers, etc. on infrastructure that consists of CPUs, Memory, Network and Storage. Accelerator hardware such GPUs are needed for emerging class of deep learning workloads with unique set of requirements that are not addressed by current container orchestration systems like Mesos, Kuberentes, and Docker swarm. In this extended abstract, we discuss the requirements to support GPUs in container management systems and describe our solutions in Kubernetes. We will conclude with a set of open issues that are yet to be addressed to fully support deep learning workloads on distributed infrastructure.Operating system (OS) allows flexible sharing and load balancing of resources like CPU, Memory and Network among multiple processes and containers. Unlike these GPU's are unique quantities (GPU 0, GPU 1, ...) and they must be allocated accordingly (e.g., allocate GPU 0 to Container 1). GPU topology, will heavily affect the bandwidth of GPU to GPU communication and must take into consideration. Moreover, GPU topology even affects GPU capabilities. In some systems, for example, GPUs on different CPU socket cannot have Peer to Peer communication capability.To address these issues, firstly, we have enabled GPU support on Kubernetes. We implemented a GPU allocator module to record GPU number-to-device mapping. Kubernetes users only request number of GPUs need for their workload; GPU allocator module maps the number to actual GPU devices according to required scheduling policy and expose the allocated GPUs to application inside the container. Secondly, we have developed two advanced GPU schedulers, a bin-packing scheduler and a topology-aware scheduler, to improve GPU utilization and GPU performance. Bin-packing scheduler tries to bundle GPU jobs to fewer servers, so that other idle servers can be reserved for potentially large jobs. Topology-aware scheduler can automatically collect GPU topology information of each worker node, and assign nodes that deliver the highest possible bandwidth to the application.Access to CPU, Memory, Network and Storage devices are abstracted by operating system (OS) application programming interface (API) calls. The OS translates the application calls into device specific calls internally. Unlike these resources, GPUs have device access calls that are not yet abstracted under OS API's so applications that require access to GPU devices need those GPU devices mounted inside the container, they need access to auxiliary device interfaces (like nvidia-uvm), and they need the GPU drivers inside that container. The device driver inside the container must exactly match the driver on the host for proper operation. To solve these issues, we enhanced Kubernetes to gather the device drivers on kubelet startup and mount these drivers into the container automatically. This ensures portability of the workloads across systems with potentially different drivers. A similar approach is taken by Mesos, Nvidia Docker and other systems.In addition, unlike CPU and memory, GPU is an external PCIe device and in our experience it experiences software and hardware failures far more frequently than the rest of the system. Failures could include bad connection to the PCIe slot, GPU kernel crash, bad power supply, and so on. To deal with such issues, we enabled GPU liveness check on Kubernetes (like the liveness check in any cloud service). The kubelet periodically checks the healthiness of GPU devices. Once the GPU failure is detected, the GPU will automatically be removed out from the resources pool.Finally to support multiple users, we added GPU quota support in Kubernetes so that GPU resources can be limited by different namespaces; We auto-labeled GPU devices model to Kubernetes worker nodes so that the job can use such information to filter GPUs. All these new features are originated from our real requirements and aim to enhance the usability of GPUs in a cloud context.We consider GPU related scheduling policy and algorithm to improve both GPU performance and utility as open issues. We plan to extend CPU topology and affinity support to Kubernetes, so that we can make CPU-GPU joint topology optimized scheduling. The exploration of CPU-GPU bandwidth will bring more possibilities for performance improvement especially for servers with NVLink technology (like IBM Minsky) between CPU-GPU.

This paper presents the containerization of virtualization method that is done to execute distributed application for smart urban farming system. Containerization is a virtualization method for executing distributed applications without the need of virtual machines. However, when dealing with container downtime, a backup solution is needed to ensure that no data lost and to ensure container is in functioning. A container downtime might be caused by connectivity failure or maximum CPU capacity utilization. Therefore, a micro services-based system was developed by leveraging on containerization technology such as Docker for its lightweight footprint in terms of infrastructure resources and container orchestration capabilities from Kubernetes to enable scalability and reliability of the overall system. The results showed that efficient data monitoring and management with the added advantage of self-healing nature of Kubernetes platform was achieved.