11results about How to "Reduce Interaction Latency" patented technology

The application relates to an intelligent voice communication system and method for real-time interaction between a device end and a large language model cloud end. It solves the problems of high interaction delay, lack of true full-duplex voice stream support, separation of audio stream and instruction execution, and poor interaction efficiency in the prior art. It comprises a device end and a cloud end. The device end comprises a microcontroller running a FreeRTOS real-time operating system, the microcontroller is connected with a signal acquisition and processing unit, and the microcontroller is connected with an NFC controller for device end interaction and communicates data with the cloud end through a wireless communication module. The cloud end comprises a full-duplex voice gateway, a GPU array, an AI acceleration card and a device perception scheduler. The application has the advantages that natural continuous conversation of voice in and out is realized; user instructions are recognized and executed in real time during the conversation process, voice feedback is returned synchronously, and low-delay, low-power-consumption, high-concurrency full-duplex voice interaction is realized.

The application provides a network test transmission optimization method and a test system, and belongs to the technical field of network testing.The network test transmission optimization method of the application transmits only the 4-layer header and load description information, compared with transmitting complete data packets, the data transmission amount of the PCIe bus is greatly reduced, and the bandwidth occupation of the PCIe bus is significantly reduced;the FPGA intelligent network card locally completes load generation, assembly and detection, does not need to rely on the CPU processing of the mainboard, reduces the data interaction delay, and improves the concurrent test capability of the test instrument.

The application discloses an IO optimization method and device based on distributed storage, equipment and a storage medium, and relates to the technical field of distributed storage. The method comprises the following steps: a distributed metadata index layer is constructed between a client and a metadata server, and a dynamic mapping relationship between a path prefix and the metadata server is maintained, so that when the client initiates a file access request, the target metadata server can be directly located based on the mapping relationship, and whether the request is routed to a temporary metadata server for load balancing is dynamically determined according to real-time load information of the target metadata server. Therefore, the problems of metadata request amplification caused by tree directory level-by-level analysis and uneven load of metadata servers caused by static directory binding in the related art are solved, and the technical effects of significantly reducing metadata interaction delay, improving the throughput of the system in the massive small file and burst IO scene, and avoiding a single metadata server from becoming a performance bottleneck are achieved.

This invention discloses a quantum-attack-resistant blockchain-based password authentication key exchange and authentication method. The method involves inputting preset parameters, a trapdoor, a matrix, a noise vector, a password, and a public key into a computer system; executing a parameter generation algorithm to generate public parameters; executing a hash key generation algorithm to generate a hash key and a projected hash key generation algorithm to generate a projected hash key; generating a tag based on the identifiers of both communicating parties and the projected hash key, encrypting it to obtain ciphertext, and transmitting and receiving the projected hash key and ciphertext from the communicating party; verifying the validity of the communicating party's projected hash key, and generating a session key based on relevant parameters; and finally, the computer system processor executes a computer program and outputs the session key. This invention integrates a lattice-based smooth projected hash function with a trapdoor and a blockchain structure, possessing quantum attack resistance, general composable security characteristics, requiring only one round of communication, and is suitable for large-scale blockchain node scenarios.

The invention relates to a message transmission method and device and electronic equipment, and belongs to the technical field of wireless communication, and the method comprises the steps: obtaining a to-be-transmitted streaming message, splitting the to-be-transmitted streaming message, and obtaining metadata and multi-mode streaming content of the to-be-transmitted streaming message; generating a first streaming message according to the metadata of the streaming message to be transmitted; sending the first streaming message to a message platform based on a session initiation protocol (SIP); and sending the multi-modal streaming content to a message platform based on the first streaming message and a hypertext transfer protocol HTTP, the multi-modal streaming content being sent to the message receiving end through the message platform. Therefore, physical separation transmission is carried out on the metadata of the to-be-transmitted streaming message and the multi-mode content, real-time transmission of the multi-mode streaming content is achieved through SIP and HTTP dual-protocol cooperation, the transmission efficiency of the multi-mode streaming content is improved, and end-to-end interaction delay is reduced.

The application relates to the technical field of data interaction, and discloses a high-efficiency dual-protocol intelligent agent data interaction system, which comprises a dual-stack deployment and channel mapping module, a cross-protocol weight distribution module, a collaborative perception logical channel construction module, a cross-protocol flow control coupling and transparent migration module and a packaging and delivery module. Dual-protocol communication components containing MCP protocol stacks and A2A protocol stacks are deployed on each intelligent agent node, and a virtual channel mapping table is established. The channel distribution weight of data packets between MCP channels and A2A channels is determined. Dual-protocol logical channels are constructed, and data packets with cross-protocol channels and high business collaboration degrees are mapped to the same logical topology path. When the dual-protocol logical channels are congested, the cross-protocol flow control coupling channel distribution weight is distributed, and protocol transparent switching is performed on the data flow on the corresponding logical topology path. The migrated data packets after adaptive packaging are transmitted to target intelligent agent nodes. The application can improve the efficiency of intelligent agent data interaction.

The utility model relates to a computing device and a computing server. The computing device comprises a universal substrate; the at least two computing acceleration cards are mounted on the universal substrate; the switch module is mounted on the universal substrate and is coupled to the at least two computing acceleration cards through data interconnection wires arranged in the universal substrate, and a data interaction path between any two computing acceleration cards in the at least two computing acceleration cards comprises the switch module; one-to-one data interconnection wiring which does not pass through the switch module does not exist between any two computing acceleration cards in the at least two computing acceleration cards. According to the method and the device, the adaptive adjustment of the interconnection relationship of the computing acceleration cards aiming at the specific computing task is realized, the optimization capability of the data interaction topology of the computing device and the computing server aiming at the computing task is improved, and the internal bandwidth between the computing acceleration cards is favorably improved; and the overall computing efficiency of the computing architecture can be improved to the greatest extent under various different computing tasks.

The invention discloses an FPGA resource scheduling method for automatic driving perception decision decoupling, and relates to the technical field of automatic driving. The method specifically comprises the following steps of: receiving multi-source heterogeneous sensor data by utilizing a reconfigurable sensory memory interface module of a dynamic logic area at the front end of an FPGA (Field Programmable Gate Array), executing time-space synchronization, and performing normalized projection to obtain a driving situation dynamic map data frame; a dual-port cache region is constructed by utilizing a cognitive fusion and working memory module of FPGA on-chip high-speed storage resources, and a multi-channel layered space-time tensor serving as an on-chip data exchange medium is maintained and refreshed in real time; a cognitive decision module of a rear-end static logic area is used for reading the dynamic map, reasoning is conducted in combination with priori knowledge called by a long-term memory interface, and a control instruction is generated; the dynamic map is defined as a standardized boundary of software and hardware interaction, and only a logic circuit of the sensory memory interface module is refreshed when the configuration of the sensor is changed. The objective of the invention is to realize perception and decision decoupling through a standardized architecture of a hardware bottom layer and improve the real-time performance and robustness of a system.

This invention discloses a device operation process optimization system and device based on interactive behavior data, belonging to the field of human-computer interaction technology. The invention performs facial recognition on a designated device for the user to be operated on, recording all identified users as potential users. If several potential users exist, user filtering is performed to determine the primary user to complete the device operation; otherwise, the operation is directly performed based on the potential user. If user filtering is performed, multiple interaction judgments are conducted using camera interaction behavior data to determine whether to execute the human-computer interaction operation, thereby identifying the primary user. During the execution of the human-computer interaction operation, a simultaneous human-computer interaction operation judgment is performed to select the issued interaction behavior command. Based on the response of the potential users to the interaction behavior command, the corresponding primary user is determined to complete the device operation. This solves the problem in existing technologies where systems struggle to accurately identify target users, leading to incorrect identity matching and potential payment errors.

This application relates to the field of human-computer interaction technology, specifically to a semantic interaction method, a semantic interaction system, a semantic interaction device, and a computer-readable storage medium based on electronic skin. The semantic interaction method based on electronic skin is executed by an edge device and includes: receiving sensor data from the electronic skin; identifying the user's original actions based on the sensor data using an action recognition model deployed on the edge device; and if the original action is successfully identified, generating natural language interaction content based on the identified original action using a semantic generation model deployed on the edge device. This solution achieves network-independent semantic interaction by locally integrating action recognition and semantic generation functions on the edge device, which helps reduce interaction latency and improve data privacy protection.

The application provides a hybrid storage implementation method, system, storage device and management device, and belongs to the technical field of data storage. The method comprises the following steps: obtaining a data write request initiated by a client; applying for storage space to a management device according to the data write request; the management device allocates a solid state disk erasure coding group space and a mechanical hard disk erasure coding group space corresponding to the data write request at one time, and establishes a binding relationship between the solid state disk erasure coding group space and the mechanical hard disk erasure coding group space; obtaining the data size of to-be-written data corresponding to the data write request; when the data size is lower than a preset threshold, writing data into the solid state disk erasure coding group space; when the data size is higher than the preset threshold, directly writing data into the mechanical hard disk erasure coding group space. The application reduces data interaction and repeated writing, and improves data write storage performance.