Parallel Sub-Plan Execution for Speech Intent Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In electronic devices using speech recognition services, processing plans with multiple end points can lead to latency due to sequential execution of actions across different end points, such as electronic devices and intelligence servers, resulting in inefficient interactions and performance delays.
Innovation Solution
A method is introduced to identify independent actions with different end points, combine them into sub-plans, and process these sub-plans in parallel, while updating the execution order based on dependencies to optimize the execution of plans with multiple end points, thereby reducing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If actions with different end points are processed sequentially in a plan, then the execution order can be maintained, but latency occurs due to waiting states
Solution Approach 1:
The patent segments the plan into multiple sub-plans based on end point groups. Actions with the same end point are grouped together into sub-plans, allowing independent parallel execution of sub-plans. This segmentation eliminates waiting states between different end points while maintaining execution order within each sub-plan, directly reducing latency and improving execution efficiency.
2Loss of time
If actions are executed sequentially across multiple end points, then dependency order is preserved, but frequent interactions between electronic device and intelligence server increase latency
Solution Approach 1:
The patent divides the plan into sub-plans grouped by end point (electronic device or intelligence server). Each sub-plan contains actions for a specific end point, reducing the frequency of interactions between different end points. Sub-plans are executed in parallel, minimizing back-and-forth interactions while preserving dependency relationships through ordered execution within each sub-plan.
3Productivity
If independent actions with different end points are processed in parallel, then latency is reduced, but system complexity increases
Solution Approach 1:
The patent segments actions into sub-plans based on end point groups, enabling parallel execution of independent sub-plans. This segmentation approach manages system complexity by organizing actions into manageable groups while achieving improved execution speed through parallel processing of these groups.
Solution Approach 2:
The patent merges actions with the same end point into unified sub-plans. This merging reduces the overall number of separate execution contexts needed while enabling parallel execution of different end point groups, thus improving productivity without proportionally increasing system complexity.
Data Source
AI summary
A method for processing plan is provided. The method includes multiple end points including obtaining an intent corresponding to a user utterance, generating the plan for setting a connection order of actions for executing the intent, determining whether there are multiple end points of the actions, determining whether the actions having different end points are independent, and combining actions having the same end points to generate a plurality of sub plans again and then processing the plurality of sub plans in parallel when the actions having the different end points are independent.