AI Computing Resource Scheduling for Low-Latency Device Collaboration

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing wireless network architectures fail to fully utilize AI capabilities of network functions like terminals and base stations, hinder AI collaboration between devices, and struggle with low-latency AI services and data privacy issues, especially when computing power resources are insufficient.

Innovation Solution

A computing power resource scheduling method where a first terminal device requests resources from a first device, such as a base station or core network, using signaling protocols like RRC or NAS, to support AI collaboration by determining and configuring necessary computing power resources.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If terminal device uses local computing power for AI collaboration, then AI service response speed is improved, but computing power resource sufficiency deteriorates

Engineering Contradiction:

ImproveAI service response speedVSAvoidcomputing power resource sufficiency

Core Design Contradiction:

SpeedVSQuantity of substance

Solution Approach 1:

The patent merges local terminal computing power with network device computing power into a unified computing resource pool. The terminal device and network device are combined to form a collaborative computing system, allowing AI tasks to be distributed and executed across both local and remote resources, thereby resolving the contradiction between fast local response and sufficient computing resources.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The network device acts as an intermediary between the terminal device and AI computing resources. It receives AI collaboration requests from the terminal, allocates appropriate computing tasks to itself or other available devices, and returns results to the terminal, thereby enabling the terminal to access sufficient computing power without requiring all resources to be locally available.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Adaptability or versatility

If NWDAF network element collects data from terminal for AI optimization, then AI service capability is improved, but data privacy protection deteriorates

Engineering Contradiction:

ImproveAI service capabilityVSAvoiddata privacy risk

Core Design Contradiction:

Adaptability or versatilityVSObject-affected harmful factors

Solution Approach 1:

The patent implements local quality by allowing different devices to retain and process different types of data locally according to their specific needs and security requirements. Each device maintains control over its own data while contributing only necessary information to AI collaboration, thereby improving AI service capability while protecting data privacy through differentiated local data handling.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12574337B2Computing power resource scheduling method and related apparatus

Publication Date: 2026.03.10 HUAWEI TECH CO LTD

AI summary

A first terminal device sends a computing power request message to the first device, where the computing power request message includes computing plane resource indication information, and the computing plane resource indication information indicates a computing power requirement of the first terminal device; and receives computing power configuration information sent by the first device, where the computing power configuration information includes first indication information and/or second indication information, the first indication information indicates a first computing power resource provided by the first device, the second indication information indicates a second computing power resource provided by the second device, and the first device is different from the second device.