17 results about "Frame aggregation" patented technology

The invention discloses a synchronization method and device for time synchronization of a network system, and relates to the technical field of computers, and the method comprises the following steps: when a master station sends a dynamic compensation synchronization frame, recording a compensation sending timestamp in a sensing identification field of a sending end; when the frame is received and transmitted from the station, the identification field is sensed at the port to determine a compensation receiving timestamp and a compensation transmitting timestamp. And the slave station identifies the dynamic compensation synchronization frame after reverse return, calculates dynamic compensation time information based on the compensation receiving timestamp and the compensation sending timestamp corresponding to the slave station, and reads and writes the dynamic compensation time information to a corresponding field. And the master station analyzes the returned dynamic compensation synchronization frame, aggregates the dynamic compensation time information of each slave station, and solves the dynamic compensation values of the master and slave stations in combination with the fixed link delay. And the master station corrects a local high-precision clock according to the data, and writes a compensation value into a local time correction register of the slave station to trigger a hardware counter to correct so as to complete synchronization of the master and slave clocks. By realizing dynamic residence separation, the problem of overlarge error is solved.

The invention provides a multi-frame fusion infrared gas target detection method and system based on knowledge distillation, and belongs to the technical field of computer vision. Inputting multiple frames of to-be-detected images into the knowledge distillation model, and extracting multi-scale initial features through multiple times of downsampling of a teacher model backbone network; inputting the same-scale initial features into the corresponding bidirectional GRU to obtain current frame time sequence fusion features; aggregated multi-scale features are obtained based on a neck feature aggregation network, and attention weighting is carried out on current frame aggregation features of the same scale by adopting other frame aggregation features to obtain enhanced features; and in the knowledge distillation stage, multi-scale features are aggregated by the teacher model to serve as supervision signals, so that the student model is aligned with teacher model feature expression on multiple scales, and a gas target detection result is obtained. Therefore, the inter-frame time sequence information is modeled for the interval video frames through the bidirectional GRU, and the multi-scale feature expression is aligned based on knowledge distillation, so that the infrared video gas target detection precision is improved.

The invention relates to a cascading energy storage communication technology, in particular to a high-voltage cascading energy storage system fault-tolerant communication method based on frame aggregation and a storage medium. The method comprises the following steps of: serially connecting and aggregating a control frame sent to a sub-controller by a main controller and control instruction frames sent to all subordinate unit bodies into an aggregation packet, and redundantly sending the aggregation packet; the aggregation packet is divided into a data packet and a synchronous packet; the sub-controller carries out three-level receiving caching and processing on the aggregated packet sent redundantly, and executes different logics on the data packet and the synchronous packet; wherein the main controller sends serial control information of each unit body to the sub-controller, which is called as high-speed downlink; and the sub-controller sends control information to each unit body, namely low-speed downlink. According to the invention, while stable operation of the system is ensured, efficient data transmission and an intelligent control strategy are realized so as to adapt to increasingly complex power system requirements.

The present invention relates to a method for constructing an aggregated frame for efficient wireless communication in a high-density environment and a wireless communication method using the same.For this, the present invention provides a wireless communication method including: generating an aggregated MPDU (A-MPDU) including a plurality of MAC protocol data unit (MPDU) groups to which independent modulation and coding schemes (MCSs) are applied, respectively; and transmitting the generated A-MPDU through the transceiver and a wireless communication terminal using the same.

The embodiment of the application provides a kind of semiconductor optical detection data processing method and device, method includes: by obtaining the multi-source detection data of semiconductor optical detection, constructs unified association key and carries out data binding to multi-source detection data, the objectification data packet obtained after binding is carried out data summary according to pre-set detection window, determines detection window data packet, executes in-frame area aggregation, window in-frame aggregation, state aggregation and hierarchical aggregation in the imaging quality index in each detection window data packet in turn, determines hierarchical state vector, and compares with pre-set reference vector, determines the degree of imaging quality recession and recession duration, and carries out correction and risk level determination to optical detection original output, determines optical detection correction output and risk level, according to risk level from pre-set strategy mapping table matching and executing corresponding scheduling action, realize optical detection data processing, the accuracy of the optical detection data and the efficiency of semiconductor quality monitoring can be improved.

Embodiments are disclosed for encrypting media access control (MAC) Header fields for Wireless LAN (WLAN) privacy enhancement. For example, a transceiver of a station (STA) or an access point (AP) can set a real time Media Access Control (MAC) header bit in a payload of an aggregated MAC Protocol Data Unit (A-MPDU) subframe to an actual value of a power management (PM) field of a MAC header of the A-MPDU subframe. The transceiver can encrypt the payload, set the PM field to an over the air (OTA) PM value, and transmit the A-MPDU subframe over the air. The OTA PM value can include all zeros, a predetermined value, or a randomized value The transceiver can also set static MAC header bits in the payload of the A-MPDU subframe to corresponding actual values of an aggregated MAC service data unit (A-MSDU) present field of the A-MPDU subframe.

Seamless client roaming for Multi-Link Device (MLD) clients may be provided. First, a Traffic Identifier (TID)-to-link map may be established by an Upper Service Access Point (U-SAP) of a multi-AP MLD entity that assigns subsets of TIDs to at least two links of the entity. For example, a client device logically associates with the U-SAP, while the client device physically connects to a first and second AP of the entity on a respective first and second link, where the first and second AP include first and second Lower Service Access Points (L-SAPs) and are non-collocated. Next, using the map, data received at the U-SAP is directed over one of the two links for transmission to the client device. Further, frame aggregation and block acknowledgment functions may be performed by one of the first or second L-SAP based on whether data transmission is over the first or second link.

Seamless client roaming for Multi-Link Device (MLD) clients may be provided. First, a Traffic Identifier (TID)-to-link map may be established by an Upper Service Access Point (U-SAP) of a multi-AP MLD entity that assigns subsets of TIDs to at least two links of the entity. For example, a client device logically associates with the U-SAP, while the client device physically connects to a first and second AP of the entity on a respective first and second link, where the first and second AP include first and second Lower Service Access Points (L-SAPs) and are non-collocated. Next, using the map, data received at the U-SAP is directed over one of the two links for transmission to the client device. Further, frame aggregation and block acknowledgment functions may be performed by one of the first or second L-SAP based on whether data transmission is over the first or second link.

The application discloses a kind of deep optimization deep-sea ecological monitoring method and system, comprising: first, obtain deep-sea image or video data and carry out pre-processing, carry out denoising and detail enhancement;Second, a kind of integrated Swin Transformer main part, BiFPN neck and target detection architecture of combined query decoder are used, high-precision, anti-interference target detection is carried out to enhanced image;Then, the structured target data detected is inputted with original image into multimodal large language model, carries out the inference and analysis of deep ecological semantics such as species attribute, behavior mode;Finally, multi-frame aggregation and community analysis are carried out to video data, and structured deep-sea ecological monitoring report is automatically generated according to detection and analysis result.The application solves the problems of small target detection difficulty, serious background noise interference, low data analysis efficiency, lack of semantic understanding and automation capability in deep-sea environment through model innovation and fusion.

The invention relates to the technical field of target tracking, in particular to an RGB-E multi-mode target tracking method and system. According to the method, a dynamic event subframe division mechanism based on event information entropy estimation is designed, the complexity of a current event flow is quantified through event information entropy, an event frame aggregation or event subframe division strategy is adaptively selected, an event feature extraction network is provided, and spatial offset parameters of pooling kernels are dynamically learned based on input event features, so that the complexity of the current event flow is quantified. Through adaptive adjustment of a local receptive field, the feature extraction process more pays more attention to effective events generated by real motion of a target, each frame of event image is subjected to 1 * 1 convolution, shunt pooling splicing and convolution to obtain first-stage features, each-stage features are equally divided into multiple sub-features, each sub-feature is divided according to different proportions along channel dimensions, and each sub-feature is divided into a first-stage feature and a second-stage feature; different pooling methods are adopted to process each part, and event features are obtained through fusion, so that the accuracy of the event features is effectively improved. According to the method, the RGB-E multi-mode target tracking precision is effectively improved.

To provide systems and methods for frame aggregation such as Medium Access Control (MAC) Service Data Unit (MSDU) aggregation.SOLUTION: A system for communication between devices includes a first device 602 that advertises a first information element indicating a first maximum size for MAC protocol data unit (MPDU) frames for the first device, selects, based at least on the first maximum size, a first polynomial for a cyclic redundancy check (CRC) for an MPDU frame received from a second device 604, 606, computes, using the selected first polynomial, a CRC for the received MPDU frame, and detects, based at least on the computed CRC, an error in the MPDU frame.SELECTED DRAWING: Figure 6

A first communication device determines that a trigger frame and another frame are to be transmitted to at least a second communication device. The first communication device determines whether the second communication device announced support of aggregation of buffer status report (BSRP) trigger frames with additional frames. In response to the first communication device determining that the second communication device announced support of aggregation of BSRP trigger frames with additional frames, the first communication device generates an aggregate media access control protocol data unit (A-MPDU) to include the BSRP trigger frame and the other frame, and transmits the A-MPDU within a packet. In response to the first communication device determining that the second communication device did not announce support of aggregation of BSRP trigger frames with additional frames, the first communication device transmits a packet having only the BSRP trigger frame.

The invention relates to a temperature control method and device of equipment, electronic equipment and a storage medium. Wherein the equipment supports data processing in a frame sequence form, the frame sequence corresponds to a frame aggregation length, and the temperature control method of the equipment comprises the following steps: detecting a temperature change rate of the equipment in a first duration in a process that the equipment performs data processing in a frame sequence corresponding to a first frame aggregation length; in response to the fact that the temperature change rate is a non-constant-temperature change rate, the first frame aggregation length is adjusted to be a second frame aggregation length, and the second frame aggregation length meets the requirement that the temperature change rate is converted into the constant-temperature change rate; and performing data processing according to a frame sequence corresponding to the second frame aggregation length so as to control the temperature change rate to be converted into the constant temperature change rate. According to the invention, hardware cost does not need to be increased, and user communication experience is optimized.

The present invention proposes a method for human behavior recognition based on a dual-stream three-dimensional convolutional neural network, which mainly solves the problems of limited feature representation caused by single-modal data and high computational complexity of cross-modal methods in the existing technology. It includes: 1) constructing a multimodal dataset; 2) constructing a separable three-dimensional residual module for decomposing the three-dimensional convolution into time and space dimensions; 3) constructing a dynamic similar frame aggregation module for learning the similarity of feature maps, grouping similar feature maps, and aggregating grouped feature maps; 4) using the modules in steps 2) and 3) to construct a skeleton stream network and an RGB stream network; 5) introducing a feature fusion module between the two stream networks, building a dual-stream three-dimensional convolutional neural network and training it; 6) inputting the behavior video into the trained network to obtain recognition results. The present invention can fully extract behavior features, reduce information redundancy and noise interference, and effectively improve the accuracy and performance of human behavior recognition.

The invention discloses a deep-sea ecological monitoring method and system based on deep optimization, and the method comprises the steps: firstly, obtaining deep-sea image or video data, carrying out the preprocessing, and carrying out the denoising and detail enhancement; secondly, performing high-precision and anti-interference target detection on the enhanced image by adopting a target detection framework integrated with a Swin Transform trunk, a BiFPN neck part and a linked query decoder; then, detected structured target data and an original image are input into a multi-modal large language model together, and reasoning and analysis of deep ecological semantics such as species attributes and behavior patterns are carried out; and finally, performing multi-frame aggregation and community analysis on the video data, and automatically generating a structured deep sea ecological monitoring report according to a detection and analysis result. Through model innovation and fusion, the problems of difficulty in small target detection, serious background noise interference, low data analysis efficiency and lack of semantic comprehension and automation capability in the deep sea environment are solved.