10 results about "Cross-layer optimization" patented technology

The application discloses a federated continual learning cross-layer optimization method for a UAV relay network, and aims at the problems of easy occurrence of catastrophic forgetting of a model and high system time delay in a streaming task training scene.The application constructs a system architecture comprising a base station, a UAV and a plurality of ground clients; a task stability index is obtained by calculating the gradient similarity between a current task and a historical task of a client, and a system efficiency index is obtained by combining the client computing time delay and the communication time delay, so as to jointly select a target client participating in aggregation; in order to anchor historical knowledge, local compensation updating is performed at the client side based on the historical task gradient, and post-aggregation compensation correction is performed at the base station side based on the historical global gradient; and a cross-layer optimization model about the client computing frequency, bandwidth allocation, UAV trajectory and relay strategy is further established to minimize the maximum completion time delay of the system. The application can improve the model learning precision, reduce catastrophic forgetting and reduce the training time delay.

The invention discloses a multi-layer unmanned aerial vehicle cooperative edge calculation method and system, which are applied to an air cooperative network consisting of an HAP and an unmanned aerial vehicle (UAV) and are used for access and unloading of ground PU and SU tasks. A ground user accesses the UAV in a CR-RSMA mode, the UAV establishes a communication link with the HAP in an OMA mode, and the HAP is selected for cooperative processing according to task delay and computing power. The AGT2 algorithm is adopted to realize double-layer stable matching of the PU, the SU and the UAV, the AGT2-HBS algorithm is combined to perform hierarchical joint optimization on power, bandwidth, task unloading proportion and calculation frequency, and the purpose is to minimize the total energy consumption of the system. The system is composed of a ground access part, a communication connection part, a cooperative calculation part, a matching optimization module, a resource allocation module and a control output module, and cross-layer optimization of multi-layer unmanned aerial vehicle cooperation can be realized.

The application provides a CKKS three-branch decision dynamic hierarchical encryption neural network training method and system. The method comprises the following steps: data encoding is performed on original input data, CKKS encryption is performed to generate ciphertext by adding a noise term to a polynomial; homomorphic calculation is performed on the ciphertext by alternating row-column encryption matrix multiplication, the intermediate features are processed in combination with an EncryptedPolyReLU activation function and a three-branch decision, encrypted features are obtained, and encrypted prediction results are finally output; the gradient is calculated by using an encrypted residual sum of squares loss function, and the Nesterov momentum method is used to accelerate the gradient update to process the encrypted prediction results, in combination with a hierarchical noise perception guide mechanism, so that the encrypted model parameters are optimized to obtain the encrypted model parameters. The application realizes efficient privacy protection training through dynamic encoding and row-column alternating encryption, and provides a safe and efficient deep learning solution for high-sensitive data in combination with three-branch decision screening and cross-layer optimization.

This invention provides a method, apparatus, system, terminal, and storage medium for 3D scene reconstruction. The method includes: responding to a rendering task execution command, acquiring the current field of view of several user terminals, and evaluating the rendering task volume within the current field of view of each user terminal; based on the rendering task volume, modeling a cross-layer optimization problem based on end-to-end latency and rendering quality constraints with the goal of minimizing system energy consumption; solving the cross-layer optimization problem to obtain optimization variables, including a target rendering model selected from a set of candidate rendering models; and based on the optimization variables, distributing the rendering task to several distributed nodes so that the distributed nodes can perform collaborative rendering to reconstruct the 3D scene. This application, by considering the differences in field of view, constructs a cross-layer optimization problem and ensures that end-to-end latency and rendering quality constraints are met, reduces the energy consumption of the rendering task and solves the resource optimization problem of distributed 3D reconstruction.

The invention discloses an intelligent storage and accurate scheduling method for a multi-energy complementary power system. The method comprises the following steps: collecting and preprocessing wind and light output, energy storage state, load and meteorological data in real time; predicting a load demand based on a hybrid neural network of LSTM and an attention mechanism; the weight coefficients of the cost and the energy abandoning rate are dynamically adjusted through a fuzzy inference system; constructing a power generation, energy storage and power grid intelligent agent collaborative decision by adopting an improved depth deterministic strategy gradient algorithm, and generating a scheduling instruction in combination with a Pareto frontier storage pool and an attention mechanism; controlling the energy storage system to charge and discharge according to the instruction; and online updating of model parameters and dynamic correction of a scheduling strategy are realized through closed-loop feedback and cross-layer optimization. According to the method, the dynamic balance problem in multi-objective optimization is effectively solved, the energy utilization efficiency is improved, the operation cost and the energy abandoning rate are reduced, the system stability is enhanced, and the energy storage life is prolonged.

Managing Low Latency, Low Loss, Scalable Throughput (L4S) traffic may be provided. Managing L4S traffic can include receiving network data comprising one or more Physical (PHY) layer metrics, one or more Media Access Control (MAC) layer metrics, and one or more network layer metrics. Network conditions are determined based on the network data. One or more L4S characteristics are then set based on the network conditions.

The invention discloses a cross-layer optimization routing protocol method and system for a high-dynamic ad hoc network scene. The method specifically comprises the following steps: firstly, establishing a cross-layer optimization model for a high-dynamic ad hoc network; then, physical layer link quality is calculated, and link quality evaluation is carried out in a windowed exponentially weighted moving average smoothing mode; counting the load condition of the MAC layer based on Kalman filtering, and carrying out updating interaction through a network maintenance message; and finally, constructing a comprehensive evaluation model, and realizing MPR selection based on a greedy algorithm for joint optimization of cross-layer resources. The system comprises a cross-layer optimization model construction module, a link quality evaluation module, an update interaction module and an MPR selection module. According to the invention, the problem of low network stability in a multi-node dynamic topology environment is solved, the end-to-end delay is reduced, the packet delivery success rate is improved, the routing overhead is reduced, and the adaptability to the dynamic environment is enhanced.

Managing Low Latency, Low Loss, Scalable Throughput (L4S) traffic may be provided. Managing L4S traffic can include receiving network data comprising one or more Physical (PHY) layer metrics, one or more Media Access Control (MAC) layer metrics, and one or more network layer metrics. Network conditions are determined based on the network data. One or more L4S characteristics are then set based on the network conditions.

The invention relates to the technical field of wireless communication and multimedia communication, and discloses a cross-protocol-layer packing optimization method for lightweight image transmission of the Internet of Things and related equipment. According to the method, based on an application layer model, an access layer model and a physical layer model, a cross-layer optimization objective function based on a weighted product form is constructed for three sub-indexes of image quality, time delay and energy consumption; constructing a cross-protocol layer model, and solving by optimizing a cross-layer optimization objective function to obtain an optimal package size and optimal transmitting power; calculating a corresponding optimal quantization step size according to the optimal package size; and the Internet of Things node encodes the acquired lightweight image according to the optimal quantization step size, and sends a generated data packet to a receiver at the optimal transmitting power. The method can be applied to the transmission of the lightweight images of the Internet of Things through the random access network, and achieves the flexible balance and collaborative optimization of the image quality, the time delay and the energy consumption.

The invention relates to the technical field of wireless communication, discloses a WiFi (Wireless Fidelity) multi-link dynamic management system and method based on AI (Artificial Intelligence) prediction and DPI (Deep Packet Inspection) technologies, and solves the problems that the existing WiFi multi-link operation technology is lack of application perception capability, link strategy static solidification, no predictive management capability and cross-layer optimization. The problems that different application differentiation requirements cannot be met, the resource allocation efficiency is low and the network performance fluctuates are solved. According to the scheme, firstly, a DPI technology is adopted to analyze a data packet in real time, the type of an application program is identified, and the bandwidth requirement, the delay sensitivity and the packet loss tolerance of the application are extracted; secondly, predicting application network requirements and future network quality changes; and then according to an application type identification result, an application network demand and a future network quality prediction result, dynamically adjusting a multi-link operation strategy, simulating an execution effect of the multi-link operation strategy through a digital twin technology, and if a simulation result meets a requirement, applying the multi-link operation strategy to an actual network.