Edge Computing Pipeline Construction for Latency Optimization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Edge computing devices face processing, memory, bandwidth, and energy constraints, which limit computing processes, and communication bandwidth issues lead to high latency when offloading processes to server devices.

Innovation Solution

A computing system with an edge computing device that selects and executes a processing stage machine learning model based on contextual data, constructing a runtime processing pipeline to generate outputs while providing metadata on processing stages.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If computing processes are executed at edge computing devices, then processing speed and latency are improved, but device resource constraints (processing, memory, energy) worsen

Engineering Contradiction:

ImprovelatencyVSAvoidresource constraints

Core Design Contradiction:

Loss of timeVSDevice complexity

Solution Approach 1:

The patent implements dynamic pipeline construction where the edge computing device selects and configures processing stages based on runtime contextual data including resource availability. The processing pipeline is not static but adapts its complexity and composition based on current device state, allowing the system to optimize between latency and resource consumption in real-time

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes operational parameters by selecting different machine learning model configurations and processing stage combinations based on resource constraints. The edge device can adjust model size, complexity, and number of processing stages according to available processing power, memory, and energy, thereby resolving the contradiction between fast local processing and limited device resources

Inventive Principle:

Principle #35Parameter changes

2Power

If computing processes are offloaded to server computing devices, then processing capability is improved, but communication latency and bandwidth requirements worsen

Engineering Contradiction:

Improveprocessing capabilityVSAvoidcommunication latency

Core Design Contradiction:

PowerVSLoss of time

Solution Approach 1:

The patent segments the computing process into multiple processing stages that can be distributed between edge and cloud. Critical time-sensitive stages are executed locally at the edge device, while less time-critical stages can be offloaded to the server. This segmentation allows the system to balance processing capability with communication latency by keeping essential processing local

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary processing actions at the edge device before potential offloading. By pre-processing data locally and only transmitting necessary results or intermediate outputs to the server, the system reduces communication bandwidth requirements and latency while still utilizing server processing capability when needed

Inventive Principle:

Principle #10Preliminary action

3Adaptability or versatility

If multiple machine learning models are stored at edge computing devices, then model selection flexibility is improved, but memory and storage requirements worsen

Engineering Contradiction:

Improvemodel selection flexibilityVSAvoidmemory and storage

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent implements a universal pipeline construction approach where a single edge computing device can dynamically configure different processing pipelines using a core set of processing stages. Instead of storing multiple complete specialized models, the system uses multi-functional processing stages that can be combined in different sequences to serve multiple purposes, reducing memory requirements while maintaining flexibility

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system dynamically selects and configures processing pipelines at runtime based on contextual data rather than statically storing multiple complete models. This dynamic approach allows the edge device to adapt its processing capabilities to different tasks using a smaller set of versatile processing stages, thereby reducing storage requirements while maintaining model selection flexibility

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS12211248B2Constructing processing pipeline at edge computing device

Publication Date: 2025.01.28 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

A computing system including an edge computing device. The edge computing device may include an edge device processor configured to receive edge device contextual data including computing resource availability data. Based at least in part on the edge device contextual data, the edge device processor may select a processing stage machine learning model of a plurality of processing stage machine learning models and construct a runtime processing pipeline of one or more runtime processing stages including the processing stage machine learning model. The edge device processor may receive a runtime input, and, at the runtime processing pipeline, generate a runtime output based at least in part on the runtime input. The edge device processor may generate runtime pipeline metadata that indicates the one or more runtime processing stages included in the runtime processing pipeline. The edge device processor may output the runtime output and the runtime pipeline metadata.