28 results about "Scalable distributed" patented technology

The invention discloses a distributed operation-oriented multi-agent collaborative recommendation method and system, and relates to the technical field of intelligent recommendation, and the method comprises the steps: configuring a private domain operation agent in each independent private domain platform, deploying a big language model-based construction demand analysis agent at a user side to receive a natural language demand description input by a user, and constructing a multi-agent collaborative recommendation system; generating recommendation task instructions of different private domain platforms; the method comprises the following steps: receiving user preference description and recommendation object description, sending the description to a corresponding private domain operation agent, calling a recommendation model to generate a recommendation result list based on the received user preference description and recommendation object description, collecting recommendation result lists returned by all private domain operation agents by a demand analysis agent, and integrating and classifying the recommendation result lists to generate a comprehensive recommendation list. According to the method, an efficient, safe and extensible distributed recommendation architecture is constructed by introducing a user demand analysis agent, a private domain operation agent and a federal learning mechanism.

A system having a horizontally scalable distributed architecture for automated firmware testing includes: test stations for testing firmware products, the test stations being divided into pools, each pool including multiple test stations; and multiple execution entities, each execution entity being configured to execute tests corresponding to an associated pool. Each competing test station sends a test initiation event to a corresponding execution entity. The corresponding execution entity receives the test initiation event from the competing test stations and executes a run test command on a selected test station among the competing test stations, causing the selected test station to execute the test based on a test sequence.

Techniques discussed herein manage backups of a service cell (SC). Each SC may include a data plane that is isolated from other SCs and comprises a distributed computing cluster (a cluster). A manifest that specifies one or more backup policies may be used to generate a full backup or a partial backup of a data set stored by the cluster. In accordance with the manifest, a signal may be sent to nodes of the cluster. In response, the nodes may transmit locally-stored data (e.g., data segments) to specified locations at a remote storage. The system may maintain a mapping of which segments correspond to data that was stored in the cluster at a time corresponding to a full or partial backup.

This application provides a distributed wireless resource scheduling method and electronic device, relating to the field of wireless communication network technology. In this method, the coordinator and computing nodes only need one round of information exchange, avoiding the huge signaling overhead and accumulated latency caused by multiple rounds of iterative negotiation. Time-consuming optimization calculations are offloaded to multiple computing units for parallel execution, achieving a linear reduction in overall computation time. Furthermore, it leverages the approximate achievable rate to provide theoretical feasibility for distributed solutions, while global coordination by edge devices ensures the performance of critical users, allowing the performance of the distributed solution to approach that of centralized algorithms, which serve as the performance ceiling. Because the coordination overhead is constant and minimal, and the computational load is effectively distributed, the system performance is far less affected by network size growth than centralized solutions. This makes it highly suitable for deployment in future network scenarios such as ultra-dense networking and large-scale carrier aggregation, enabling high-performance, low-overhead, and scalable distributed resource allocation under an O-RAN architecture.

A first hypervisor running on a first processor cluster is provided. During operation, the first hypervisor can determine a first set of processing nodes and a first memory unit of the first processor cluster in response to the booting up of a first Basic Input / Output System (BIOS) of the first processor cluster. The first hypervisor can discover a second hypervisor running on a second processor cluster comprising a second set of processing nodes and a second memory unit. The first hypervisor can operate, with the second hypervisor, a distributed system comprising the first and second sets of processing nodes and the first and second memory units. The first hypervisor can then operate, with the second hypervisor, a global virtual machine on the distributed system. The virtual memory space of the global virtual machine can be mapped to respective memory spaces of the first and second processor clusters.

The present application discloses a multi-source action data management system and method under an extensible distributed storage architecture, which relates to the field of data management technology. The method includes accessing the original collected data information collected by the data acquisition module based on a multi-source data interface and outputting a data access completion signal; compressing the original collected data information to obtain multi-source compressed data information, calibrating the timestamp of the multi-source compressed data information, and marking the collection parameters of each data in the multi-source compressed data information; obtaining data dynamic sharding strategy information according to the data type and collection time of the multi-source compressed data information, judging the health of each storage node in the storage cluster, screening effective storage nodes, and distributing each data to the effective storage nodes based on the data dynamic sharding strategy information, judging whether the storage cluster needs to be expanded, and if expansion is required, expanding based on the data migration algorithm. The present application has the effect of improving the efficiency of multi-source action data management.

An integrated circuit (IC) design may be received by a first process. First reset domain crossing (RDC) analysis may be performed by the first process on the IC design to generate first RDC analysis results. A set of processes may be spawned by the first process, where a second process in the set of processes may read the first RDC analysis results, obtain information for an RDC scenario from the first process, perform second RDC analysis on the IC design based on the information for the RDC scenario to obtain second RDC analysis results, and send the second RDC analysis results to the first process. The second RDC analysis results received from the set of processes may be merged by the first process to obtain merged RDC analysis results.

The invention provides a binary sensor network extensible distributed estimation method and system based on an energy harvesting and dynamic event triggering protocol, and belongs to the field of sensor network control. The problem that the service life of a sensor network node is short due to the fact that an existing sensor node is limited in data communication energy is solved. Comprising the following steps: establishing a linear dynamic model based on binary sensor network measurement; obtaining useful measurement information based on the difference between the true value and the estimated value of the binary sensor; a dynamic event triggering protocol and an energy harvester are introduced to schedule transmission of neighbor node information; constructing a distributed estimator for each binary sensor, and giving an expected finite time domain performance constraint; sufficient conditions are established for the presence of a desired distributed estimator. The communication cost is reduced, and the system reliability is improved; the communication burden is relieved, and the probability of incomplete information is reduced; the problem of insufficient energy during communication between the nodes is solved, and the stability of the system is enhanced.

The invention discloses an unmanned aerial vehicle-based urban live-action three-dimensional model production method. The method comprises the steps of data acquisition, three-dimensional reconstruction, big data storage, three-dimensional model visualization and urban data protection. The invention belongs to the technical field of urban three-dimensional modeling, and particularly relates to an unmanned aerial vehicle-based urban live-action three-dimensional model production method, which adopts urban live-action three-dimensional reconstruction, solves the problems in a multi-view image through a self-adaptive multi-view aggregation and depth estimation method, and improves the production efficiency of an urban live-action three-dimensional model. The precision and the efficiency of urban three-dimensional reconstruction are improved by using paired view weighting, spatial pyramid pooling and recursive regularization modules through a differentiable warping method and weighting cost aggregation; urban data storage is adopted, storage resource allocation and scheduling are optimized in a large-scale distributed data storage system through a dual consensus algorithm, local consensus and global consensus are combined, conical constraints are introduced, and an efficient and extensible distributed data storage optimization scheme is provided.

A system and method for self-executing graphs wherein execution semantics are encoded within graph elements themselves rather than imposed by external schedulers. The system comprises a dynamic property graph with vertices and edges that encode execution semantics specifying computational operations and graph-internal triggering conditions. An execution engine evaluates these triggering conditions by monitoring graph state, detects satisfaction through graph-internal evaluation, and initiates bound computational operations. Triggering conditions are expressed in terms of vertex or edge traversal, property changes, geometric properties such as curvature, or topological connectivity patterns. The graph autonomously determines when and which operations execute based on graph-resident execution semantics. In distributed embodiments, multiple local coordinators evaluate triggers within assigned graph regions and coordinate through peer-to-peer messaging without centralized scheduling. Applications include autonomous cognitive systems, adaptive control systems, and scalable distributed computation requiring self-regulating execution without centralized bottlenecks.

The application provides a container image scanning method and system based on an elastically scalable distributed load balancing, and belongs to the field of container image security, and comprises the following steps: S1, a Server end acquires a container group template based on Kubernetes admission control; S2, the Server executes different scanning strategies according to the template; S3, an Agent acquires an image layer from an image warehouse according to an image layer distributed by the Server, scans the image layer, and returns the scanning result to the Server; S4, a Job performs a scanning task on a node according to an image name, and reports a scanning result; after the task is executed, the Job container is automatically destroyed; S5, the Server distributes a scanning task according to a resource condition of the Agent, and dynamically adjusts the number of Agents according to a load; and S6, the system checks the uniqueness of an image layer ID, and avoids repeated scanning. The method realizes efficient container image scanning and resource management.

This invention discloses a scalable distributed multi-chamber open-path photosynthetic measurement system and method, relating to the field of plant physiological and ecological measurement technology. The system includes a main unit and at least two distributed chambers. The main unit centrally manages gas supply, concentration mixing, flow control, and high-precision concentration differential analysis. Each chamber is an independent unit, equipped with an adjustable LED light source, thermoelectric temperature control module, multiple types of sensors, independent power supply, and wireless communication module. An open path is formed through the gas supply pipeline and the return pipeline. The core of the method lies in the main unit polling and sampling the return gas from each chamber according to a preset program through a multi-channel automatic switching valve group. Combined with a "flushing stabilization" strategy, a single gas analysis unit is used to achieve high-precision, high-throughput parallel measurement of multiple chambers. This invention solves the problems of low throughput and high cost of traditional open-path instruments, achieving a balance between high precision and high throughput, and supports in-situ deployment in the field and remote monitoring.

The invention discloses a safe extensible distributed system implementation method supporting corruption self-repairing, which comprises the following steps: a registration node generates a key pair and pays guarantee deposit, and a node passing verification enters a candidate pool; each fragment runs a Byzantine fault-tolerant consensus algorithm to process the transaction, and a state snapshot is broadcasted when the time is ended; when activity loss is detected, the node broadcasts a stagnation report and reports in a cross-piece manner after a threshold value is reached, and other nodes verify the authenticity of the report; when security loss is detected, the nodes broadcast conflict state proof, and the nodes in the fragments broadcast committed snapshots and receipts; if no security loss exists, selecting the highest committed height state, otherwise, selecting the bifurcation state with the highest total transaction amount; comparing the signature with the receipt to identify malicious nodes; new nodes are selected from the candidate pool to replace the old nodes, and all the nodes committ new states and member configurations. The method and the device are used for solving the problems that a fragment configuration scheme of an existing fragment block chain is poor in expandability and low in security, and too high extra overhead is introduced for transaction processing.

The application discloses a kind of turbo-shaft turboprop engine mechanical hydraulic product test intelligent knowledge fusion platform.The platform includes distributed mass test data storage, data warehouse-text knowledge fusion and system plug-in function enhancement mechanism;Based on distributed mass data storage, flexible and scalable distributed storage scheme is provided for the mass data generated by test;On this basis, mass data is organized through data warehouse, based on data warehouse-text knowledge fusion, unstructured text data and structured mass data stored in data warehouse are linked through the form of URL, the fusion of unstructured and structured is realized;Through system plug-in function enhancement mechanism, the function of adding and updating decision algorithm at any time is realized.The application solves the technical problems of storage, management and query of product mass test data, provides data support for decision algorithm, realizes the extension and dynamic updating of system function.

One embodiment provides for a method of transmitting data between multiple compute nodes of a distributed compute system, the method comprising creating a global view of communication operations to be performed between the multiple compute nodes of the distributed compute system, the global view created using information specific to a machine learning model associated with the distributed compute system; using the global view to determine a communication cost of the communication operations; and automatically determining a number of network endpoints for use in transmitting the data between the multiple compute nodes of the distributed compute system.

Computer-implemented method (600) for processing sensor data by a first endpoint device (102), comprising: Transmitting (604) a first message to other endpoint devices (102) in a network (130), the first message indicating that the first endpoint device (102) is providing a first service; Receiving respective second messages from a plurality of endpoint devices (102), wherein the respective second messages indicate that the plurality of endpoint devices (102) also provides the first service; Forming (606) a private communication network with the multitude of endpoint devices (102); Receiving (616) sensor data via a sensor; Performing (618) a first operation to extract a first feature from the sensor data; and Transmitting (620) a third message containing data associated with the first feature to the plurality of endpoint devices (102).

This invention discloses a highly scalable serverless computing distributed scheduling system, comprising a gateway, a central layer, and a local layer. The central layer further includes an auto-scaling engine and a global scheduler. The local layer includes a local scheduler on each node. The auto-scaling engine processes user function call requests from the gateway and determines whether to generate a scaling task. The local scheduler interacts with received scaling tasks through a local check phase, a local probing phase, and a global scheduling phase. The global scheduler schedules scaling tasks issued during the global scheduling phase by finding the optimal node among all nodes in the cluster. This invention combines "local-central" scheduling with highly scalable distributed instance scheduling technology based on local probing, significantly improving the instance scheduling throughput and elastic scaling capabilities of serverless platforms.

Embodiments described herein relate to computer systems and methods for cognitive tests for scalable precision education that involves artificial intelligence, natural language processing, machine learning, large language models, model training, and scalable distributed computing infrastructure. Embodiments described herein relate to computer systems for cognitive tests for scalable precision education for a user. The system can classify and extract skill items and knowledge items from the one or more databanks of items based on categorization confidence scores generated by tuned large language models. The system can use conversation agents enabled by the one or more large language models for customized prompting based on user history records and user evaluations in skill and knowledge. The system generates a prescribed curriculum customized for users.

Scalable distributed adversarial training techniques for robust deep neural networks are provided. In one aspect, a method for adversarial training of a deep neural network-based model by distributed computing machines M includes, by distributed computing machines M: obtaining adversarial perturbation-modified training examples for samples in a local dataset D(i); computing gradients of a local cost function ƒi with respect to parameters θ of the deep neural network-based model using the adversarial perturbation-modified training examples; transmitting the gradients of the local cost function ƒi to a server which aggregates the gradients of the local cost function ƒi and transmits an aggregated gradient to the distributed computing machines M; and updating the parameters θ of the deep neural network-based model stored at each of the distributed computing machines M based on the aggregated gradient received from the server. A method for distributed adversarial training of a deep neural network-based model by the server is also provided.

This invention relates to the field of hard disk technology, and solves the technical problems of low global storage utilization in existing technologies, unnecessary redundancy waste when node reliability is high, and inability to provide sufficient fault tolerance when node reliability is insufficient. In particular, it relates to a data storage system, method, and storage medium based on edge computing. The method involves the following steps: collecting the raw data stream of the target data device, and obtaining preprocessed data blocks through data preprocessing based on the raw data stream. This invention uses consistent hashing to map nodes and data blocks to a ring space, providing a structured and scalable distributed basis for replica placement. It calculates the data reliability benefits under different replica numbers through an exponential decay availability model, and solves the Nash equilibrium point through a particle swarm optimization algorithm, minimizing global storage costs while ensuring availability, effectively solving the problem of excessive data redundancy between edge nodes.

A scalable, distributed load control system for home automation based on a network of microphones may include control devices (e.g., load control devices) that may include microphones for monitoring the system and communicating audio data to a cloud server for processing. The control devices of the load control system may receive a single voice command and may be configured to choose one of the load control devices to transmit the voice command to the cloud server. The load control devices may be configured to receive a voice command, control a connected load according to the voice command if the voice command is a validated command, and transmit the voice command to a voice service in the cloud if the voice command is not a validated command. The voice service to which the load control devices transmit audio data to may be selectable.

The invention discloses an SCADA method and system based on MQTT edge-to-service, and belongs to the technical field of industrial Internet of Things. The system adopts a five-layer architecture of an equipment layer, an edge layer, a centralized management and control layer, a development layer and a security isolation layer. The method is characterized in that edge nodes collect multi-protocol industrial equipment data at high frequency; the method comprises the following steps: integrating MQTTBroker and sFTP services by an edge controller deployed in a DMZ area to realize data aggregation, edge calculation and offline caching, and uploading data to a central server through independent dual channels; and the central server carries out centralized monitoring, analysis and instruction issuing. And the development layer realizes configuration multiplexing through a unified information model. The safety level is combined with DMZ isolation, double firewalls and transmission encryption to construct deep protection. The system solves the problems that a traditional SCADA system is difficult to expand, poor in heterogeneous compatibility, low in remote operation and maintenance efficiency and the like, and distributed monitoring which is low in cost, high in safety and easy to expand is achieved.

The invention belongs to the field of data processing, and particularly relates to an analog machine modular distributed interface and multi-bus fusion system and a data processing method. The method aims at solving the problems that development and maintenance cost is high, system reusability is low, and a standardization process is hindered. The system comprises an equipment management database, a data reading and processing module, a modular processing module, a distributed interface module and a multi-bus fusion platform. The system processes equipment information in real time through dynamic memory allocation and a multi-layer nested data structure, adopts an independent function modular design, and integrates various bus / Ethernet / serial port interfaces. The flight simulation platform realizes dynamic binding of equipment channels and simulation variables through shared memory preloading, and realizes efficient communication with analog machine electronic equipment in combination with hardware acquisition equipment, so that a modular extensible distributed interface system is formed. According to the invention, efficient expansion and real-time processing are realized through modular general design, the customization cost is reduced, and the standardization is improved.

Live media streams are delineated for distributed encoding and fragmentation in a dynamically scalable distributed resource system. In some examples, live MPEG-2 media streams are separated into groups of pictures (GOPs) and converted into jobs for encoding and fragmentation systems. Multiple jobs may be created for the same live media stream to provide for different quality levels for each fragment of each channel. Shared resources such as a pool of dynamically scalable virtual machines can be used to process the individual jobs. Encoded fragments may be H.264 fragments maintained on shared storage. A particular live stream corresponding to a particular quality level can be reconstructed using the encoded fragments.

One embodiment provides for a non-transitory machine readable medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform operations comprising providing an interface to define a neural network using machine-learning domain specific terminology, wherein the interface enables selection of a neural network topology and abstracts low-level communication details of distributed training of the neural network.

The invention discloses an extensible distributed financial system and method based on post-quantum security and auditing and a storage medium. A market transaction method of the system comprises the steps that a network node receives a transaction request from a client; performing simulation execution on the transaction request by a peer-to-peer node in the network node, and signing an endorsement response by adopting a password agility module; the sorting service of the system adopts a consensus mechanism which separates transaction data propagation from a Byzantine fault-tolerant consensus of transaction metadata certification, globally sorts endorsed transactions, and generates blocks containing ordered transactions; verifying the validity of each transaction in the block through a zero-knowledge proof protocol; and providing an auditing key for an auditing mechanism to realize supervision and auditing. According to the invention, the strong guarantee requirement of the future digital financial market on the long-term safety of the system, the extremely high performance requirement on transaction processing, the strong protection requirement on transaction privacy and the strong penetration requirement on national financial supervision can be met at the same time.

A system for dynamic resource allocation in high-performance computing environments, comprising: a monitoring module for collecting real-time performance metrics from distributed compute nodes; a predictive analytics engine that uses deep learning models to predict resource requirements; a reinforcement learning controller that dynamically allocates resources based on workload variations; a decentralized orchestration layer to ensure fault tolerance and scalability; intelligent load balancing to reallocate workloads based on real-time insights; and an energy optimization module to reduce power consumption.

One embodiment provides for a method of transmitting data between multiple compute nodes of a distributed compute system, the method comprising creating a global view of communication operations to be performed between the multiple compute nodes of the distributed compute system, the global view created using information specific to a machine learning model associated with the distributed compute system; using the global view to determine a communication cost of the communication operations; and automatically determining a number of network endpoints for use in transmitting the data between the multiple compute nodes of the distributed compute system.