19 results about "Network size" patented technology

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.

The application discloses a scalable network security risk checking method, and relates to the technical field of network security, and comprises the following steps: establishing a network information collection mechanism to collect network information from each node of a target network; establishing a risk feature library, initializing the feature library, storing feature information of common network security threats in the risk feature library in a classified manner, and establishing indexes and classification labels; extracting a network topology graph from network topology information, calculating a network scale evaluation value according to the network topology graph; and generating a checking strategy according to the network scale evaluation value. The application flexibly generates a checking strategy according to a network scale evaluation value, is suitable for networks of different scales and architectures, can timely adjust the checking strategy when the network scale changes, ensures the effectiveness of risk checking, and effectively overcomes the limitations of traditional methods designed for specific network architectures.

The application belongs to the field of FPGA-based hardware accelerator design, and proposes a point cloud target detection neural network accelerator and an acceleration method based on FPGA. The convolution accelerator comprises an instruction decoding module, a convolution module, an accumulation module, a truncation module and an activation function. The intermediate value cache only temporarily stores the channels of one feature map, consumes less BRAM, and is suitable for accelerating the point cloud target detection neural network with large feature map size, large number of channels and large network size. The application has flexible configurability and excellent reusability, so the application can accelerate different convolution layers of the point cloud target detection neural network, and is suitable for being integrated as an IP into a large design, reducing FPGA resource consumption and improving resource utilization. Moreover, the weight and the input feature map share one data input channel, and the convolution intermediate value is temporarily stored on-chip BRAM, reducing the feature map carrying frequency and saving the FPGA interface bandwidth.

The invention discloses a TDMA (Time Division Multiple Access) ad hoc network access method based on a dynamic time slot frame structure, which comprises the following steps: on the basis of networking in a fixed TDMA mode, interacting an on-line node mapping table and an off-line node mapping table through signaling to obtain the number of on-line nodes and node IDs (Identity) of the whole network so as to adjust the design of an on-network time slot frame structure, reduce the occupation of useless time slot resources and improve the access efficiency of the whole network. The frame efficiency is effectively improved, and the service transmission quality is improved. According to the method, the time slot frame structure can be adjusted according to the number of online nodes, the node information is notified to the whole network in a mapping table mode, the signaling overhead is reduced, the method adapts to different network scale scenes through the access mode of the dynamic time slot frame structure, and the service transmission quality is effectively improved.

The embodiment of the invention provides a network reliability analysis method based on a mixed K / Q-cascade model. The network reliability analysis method is based on a mixed K / Q-cascade model, and comprises the following steps: 1, initializing model parameters; step 2, generating an initial random network and extracting a node initial degree; step 3, first-round failure detection is carried out, and the first-round failure detection is carried out only for k-cascade nodes; 4, recording the initial change of the network scale; step 5, cascade failure iteration judgment and network updating: carrying out loop iteration until no node fails; and step 6, visualizing network scale evolution. According to the hybrid K / Q cascade model provided by the invention, the local dependence and heterogeneity of network nodes are considered, and the node difference is also considered. Analysis is carried out according to a mixed K / Q cascade model, and besides information rapid propagation and information rapid stop, information propagation also has a third state, namely a metastable state. When the complex network is in the state, information spreading almost stops for a long period of time in the earlier stage until sudden collapse occurs at a certain moment in the middle. The phase change phenomenon is more suitable for some actual scenes.

The embodiment of the invention provides a cooperative task complexity analysis method and system based on task network characteristics, and the method comprises the steps: carrying out the hierarchical decomposition of a command and control task into atomic tasks according to four dimensions, constructing a task network model, traversing the task network model, extracting network scale characteristics, structural relation characteristics and uncertainty characteristics, and carrying out the calculation of the complexity of the atomic tasks. Complexity calculation is carried out on the network scale feature, the structural relationship feature and the uncertainty feature, a corresponding complexity quantification result is determined, and optimization adjustment is carried out on the task network model according to the complexity quantification result, so that the task network complexity of the task network model is reduced, and the task network complexity of the task network model is improved. The accuracy and efficiency of command and control task complexity measurement can be improved.

According to the routing type CGRA fault-tolerant simulation platform, when the simulation platform runs, the instruction register sends an operation code to the PE for configuration, the PE loads a data packet from the data register, and the PE completes corresponding operation and outputs an operation result to the local router or the data register; the local router sends the data to the target PE node through a multi-hop route to complete calculation and output a result; wherein the data packet contains operands and address information. The platform realizes a routing type CGRA interconnection structure model, a user can flexibly configure all PE, network scale, routing algorithm and other parameters, and can read a data packet from any specified PE node and forward the data packet to a target PE through multi-hop routing so as to simulate time delays under different conditions.

The invention provides an asynchronous network search method, system and device for a frequency hopping ad hoc network and a medium. The method comprises the following steps: determining a network scale M of the frequency hopping ad hoc network and a time error N between network nodes; determining a corresponding synchronous frequency point Ftod in combination with the original data key PK and a preset operation rule, and returning to the first N-1 minutes to obtain N synchronous frequency points; uniformly setting N + 1 synchronous time slots in each subframe through the time error N, and selecting a continuous time slot between two synchronous time slots as an asynchronous network detection time slot; and based on the asynchronous network detection time slot, sending and receiving of broadcast messages are carried out by using the synchronous frequency point, and asynchronous network discovery and time synchronization are completed. According to the method, through time error and synchronization frequency point design, efficient network discovery and time synchronization are realized, different scene requirements can be adapted, time delay fluctuation is reduced through a detection and backspacing mechanism, and network real-time performance and stability are improved.

The invention relates to a networking control method and device, equipment, a storage medium and a product. A current popularity threshold value, a target networking magnitude and a current actual networking magnitude corresponding to a live broadcast stream are obtained, the popularity threshold value is used for determining whether to perform networking on the live broadcast stream based on the popularity of the live broadcast stream, and the networking magnitude is used for representing the carrying capacity of a constructed network on the live broadcast stream. And further, under the condition that the target networking magnitude and the current actual networking magnitude meet a preset adjustment condition, the current heat threshold is adjusted, a target heat threshold is generated, and the target heat threshold is used for regulating and controlling the networking magnitude of the networking for processing the live broadcast stream. Compared with the prior art, the embodiment of the invention has the advantages that the heat threshold can be adjusted in real time, the networking structure can be optimized in real time, the situation that the networking magnitude is insufficient or excessive is reduced, the networking magnitude is kept stable, the cost income is not unbalanced, and the income is maximized.

This invention provides a method for dynamic resource allocation in heterogeneous networks based on graph attention. The method involves selecting several access points for each user in the heterogeneous network according to a first parameter; constructing a directed graph for parallel transmission in the heterogeneous network based on each user and all access points selected by each user, and defining the feature vector of each node in the directed graph; weighting and aggregating the feature vectors of each node in a single-layer graph attention layer of the graph attention network using a multi-head graph attention mechanism to obtain the first node feature; iteratively updating the first node feature in multiple layers of the graph attention network to obtain resource allocation coefficients; and dynamically allocating network resources based on these resource allocation coefficients. Compared to traditional optimization methods and conventional DNNs, this method achieves flexible adaptation to different network scales and user numbers, improving the service quality and resource utilization efficiency of parallel transmission in 6G multi-band heterogeneous networks.

The application discloses an implementation method and device for realizing localization of file transmission intelligent support, and relates to the technical field of artificial intelligence. The method comprises the following steps: determining the network scale of an AI model according to the hardware constraints of a to-be-deployed system; selecting an AI model smaller than the network scale, and preparing a specific data set related to file transmission; training the AI model by taking the specific data set related to file transmission as a data sample, to obtain a basic model; fine-tuning a discriminant model and a plurality of expert models from the basic model through fine-tuning technology, the discriminant model being used for matching an input question to a corresponding expert model, and the expert model being used for obtaining a result based on the input question; and publishing the discriminant model, the plurality of expert models and file transmission software to the to-be-deployed system. The embodiments provided by the application are applied to servers with hardware condition restrictions and network access restrictions, realize a faster reasoning speed, and improve the intelligence of file transmission.

The present application relates to the technical field of wireless self-organizing network, in particular to a mutual synchronization neighbor discovery method of TDMA wireless self-organizing network, which realizes time synchronization of multi-hop neighbor nodes in the absence of Beidou clock synchronization signal through time slot structure, synchronization algorithm, neighbor discovery, best sector selection and network scale adjustment steps, and through the way of competing for master node, the present application ensures unique master clock of the whole network and realizes fast access of new nodes, thereby ensuring stable and reliable communication channel quality, and at the same time, the present application counts the number of actual nodes in the network and adjusts the network node scale according to the number, thus solving the shortcomings of the existing self-organizing networking method.

The application provides an RDMA network data processing method and system based on driving middleware, and the method is applied to a network card driver, and the method comprises the following steps: obtaining a request of an RC type connection created by an application, constructing a request of a UD type connection based on the request of the RC type connection; obtaining a WQE of the RC type issued by the application, and cutting a large message in the WQE into a plurality of small messages with an MTU granularity; delivering the WQE to a connection of the UD type which is in an RTS state, adding a number to each small message in sequence, adding a number and an identification field to each ordinary message or small message in sequence, and sending to a destination; obtaining feedback information of the destination, determining whether data retransmission is needed based on the feedback information, and performing data retransmission on the corresponding ordinary message or small message. The scheme is applied to a network card driver program, a request of a UD type connection is constructed based on a request of an RC type connection, a large QP does not need to be constructed for the UD type connection, and the stability of network scale expansion is ensured.

The application provides a wiring method, system and equipment of a data center standardized network and a storage medium, relates to the technical field of network wiring, and the wiring method of the data center standardized network comprises the following steps: acquiring the network scale of a network to be wired, the first positions of respective core switches in the network to be wired and the second positions of respective access switches in the network to be wired; determining the connection relationship of the respective core switches and the respective access switches; generating respective labels of respective network lines in the network to be wired according to the network scale and the connection relationship, wherein the respective network lines are used to connect the respective core switches to the respective access switches according to the connection relationship; and generating a pre-wiring scheme of the respective target network lines corresponding to the respective labels in the respective network lines according to the corresponding target first positions of the respective labels in the respective first positions and the corresponding second target positions of the respective labels in the respective second positions. The application can simplify the labels of the network lines and reduce the labor and time costs during wiring.

The application discloses a time division multiple access multiplexing method for mesh networking and a wireless ad hoc network communication system. In the method, the network works in a time division mode, the network comprises N nodes, only one or i nodes are designated to receive in a time slot, and other nodes transmit, each node has i beam directional transmission and j beam directional receiving capabilities, wherein (1) when the network scale is small, i.e. N<=8, i=1 and j=N-1 are taken, each node receives simultaneous frequency signals from other N-1 nodes in turn according to a time slot; (2) when the network scale is medium, i.e. 8<N<=32, i=1 and j=(N-1) / M are taken, M is the number of transmission node groups, at this time, one node is designated as a receiving node, other nodes are divided into M groups as transmission nodes, each group occupies a time slot, the nodes in the group simultaneously transmit frequency signals to the receiving node, and after M time slots, the transmission of the whole network to the first receiving node is completed, then a second receiving node is changed, and the process is sequentially repeated, and after M*N time slots, one network transmission period is completed; and (3) when the network scale is large, i.e. 32<N<=128, i>=2 and j=(N / i-1) / M are taken, the network is divided into i sub-networks, time domain synchronization and frequency domain orthogonality are kept between the sub-networks, and the sub-networks work in the mode in (2).

The application provides a hybrid access control method and a manned-unmanned flexible self-organizing network system based on a multi-channel access structure; the method comprises the following steps: in an initialization stage, configuring a plurality of service queues and buffer spaces thereof; synchronously processing each subnet entering the network; adopting a combination of a network-wide broadcast time period, a multi-user polling time period and a multi-user contention time period to perform multi-service queue scheduling in a time period and a time slot; and the system implements the method in a layered and clustered flexible self-organizing network through the established multi-channel access structure. The application adopts a layered and clustered network architecture and a multi-channel access structure, realizes frequency division group networking and flexible network control of each cluster, adopts a hybrid access control method to realize service classification and scheduling, solves the problems of limited network scale and capacity bottleneck in the existing self-organizing network, and meets the requirements of large-scale networking, flexible clustering and partitioned service exchange of manned-unmanned cluster communication.

This application relates to a multi-level hierarchical networking method and apparatus based on the OpenHarmony soft bus. This embodiment achieves effective expansion of network scale through a multi-level hierarchical architecture. The original connection limit of the OpenHarmony soft bus is applied to internal area management. The first central node of each area still follows this limit, but through hierarchical management, multiple such areas can be supported. By increasing the number of levels, a larger-scale network deployment can be achieved, providing a highly scalable solution for large-scale IoT applications. Moreover, the process of any leaf node discovering another leaf node first involves querying the local cache. If the local cache does not contain the node, it queries the cache of the first central node in the corresponding network area. If it is still not found, the other leaf node may be a leaf node in a different area, which reduces query response time and improves the overall stability of the network.

This invention discloses a lightweight face recognition method and system based on SqueezeNet, belonging to the field of face recognition technology. The technical problem this invention aims to solve is how to reduce the network size and achieve face recognition through a more compact network. The technical solution adopted is as follows: The SqueezeNet network is improved; the improved SqueezeNet network is pre-trained on the MS-Celeb-1M dataset for 100k training rounds; 25% of the data from the VGGFace2 training set is randomly selected for iterative pruning of the network; the pruned network is trained and evaluated using 90% of the data from the VGGFace2 training set, and the resulting network is fine-tuned to recover the potential accuracy loss caused by filter removal; the remaining 10% of the data from the VGGFace2 training set is used for validation; and the pruned network is deployed on a mobile device. The system includes an improvement module, a pre-training module, an iterative pruning module, a fine-tuning module, and a validation module.

The application provides an optical remote sensing target detection lightening method based on a knowledge distillation network, adopts a model compression method of knowledge distillation, reduces the network size, generates a lightening child network for target detection conforming to on-board load deployment, namely, on the basis of a general deep learning framework, a reasonable lightening network structure is designed according to a limited on-board GPU of a computing resource, so that the model keeps a high detection accuracy, and fast, accurate and flexible remote sensing image target detection is realized, on-orbit real-time calculation of key frame remote sensing images is realized, and the method can be applied to an on-board GPU platform with limited computing resources.