AI Model Transmission Over Air Interface With Bearer Configuration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current 3GPP protocol is inadequate for efficiently transmitting large AI/ML models between wireless communication entities due to size, quality, and delay constraints, leading to potential performance degradation or failure of AI/ML applications.
Innovation Solution
A method and apparatus for transmitting and receiving AI/ML models by configuring terminal equipment to send parts of the model based on interaction information, utilizing signaling radio bearers and data radio bearers to optimize transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the current 3GPP protocol is used for model transmission, then the transmission process is simple, but the transmission quality and timeliness cannot be guaranteed for large models
Solution Approach 1:
The patent segments the model transmission process into multiple phases: configuration phase (where transmission parameters are configured), transmission phase (where the model is actually transmitted), and completion phase (where acknowledgment is received). This segmentation allows for quality control at each stage while maintaining overall process manageability, resolving the contradiction between transmission quality and process complexity.
Solution Approach 2:
The patent implements preliminary configuration actions before model transmission, including setting transmission parameters, selecting bearers, and establishing quality requirements. This preliminary action ensures that transmission quality is guaranteed from the outset, while the structured preparation reduces complexity during the actual transmission by having all parameters pre-defined.
2Loss of time
If the current 3GPP protocol is used for model transmission, then the implementation is straightforward, but the delay control cannot be met
Solution Approach 1:
The patent introduces dynamic parameters for delay control, including maximum transmission delay, delay budget, and timing requirements. These dynamic parameters allow the system to adapt delay control to specific transmission scenarios, achieving timely delivery without requiring overly complex static control mechanisms. The delay control adjusts based on real-time transmission conditions.
Solution Approach 2:
The patent implements feedback mechanisms where the receiving entity sends completion indications back to the transmitting entity, confirming successful reception. This feedback loop allows for delay measurement and control, ensuring timeliness requirements are met while maintaining manageable complexity through standardized feedback protocols.
3Reliability
If the current 3GPP protocol is used for model transmission, then the protocol simplicity is maintained, but the integrity of model data cannot be ensured
Solution Approach 1:
The patent implements error protection mechanisms in advance, including integrity verification and error handling procedures configured before transmission. This beforehand cushioning ensures data integrity is protected from potential errors, while the structured error handling prevents complexity from escalating during actual transmission by having predefined recovery procedures.
4Reliability
If larger models are transmitted to achieve better AI performance, then the model capability is improved, but the transmission resource consumption increases
Solution Approach 1:
The patent changes transmission parameters including model size, compression level, and transmission format to optimize the balance between AI performance and resource consumption. By adjusting these parameters, the system can transmit sufficiently large models for good AI performance while controlling data volume through compression and efficient encoding, resolving the contradiction between model capability and resource usage.
Data Source
AI summary
A model transmission apparatus, configured in a terminal equipment, includes: a receiver configured to receive configuration information from a network device, the configuration information being used for configuring or indicating the terminal equipment to transmit at least a part of a model stored in the terminal equipment; and a transmitter configured to transmit the at least a part of the model to the network device according to the configuration information.