Edge-WTRU DNN Segmentation for Cloud Gaming Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cloud gaming architectures face challenges in meeting low-latency requirements due to high network latency, which affects the user experience by introducing delays that can make games unplayable, especially in fast-paced games where delays above 60 ms can be perceived and those above 100 ms render first-person shooter games unplayable.
Innovation Solution
A novel architecture that distributes a Deep Neural Network (DNN) model between the Wireless Transmit/Receive Unit (WTRU) and an Edge device, enabling online, in-situ learning and prediction of user actions, allowing for low-quality video rendering on the WTRU and high-quality prediction by the Edge device, or vice versa, to optimize latency and improve gaming experience.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If cloud gaming is implemented with centralized processing, then video quality can be high, but network latency increases making games unplayable
Solution Approach 1:
The patent segments the neural network model into multiple parts distributed across different devices. The WTRU holds a first portion of the DNN model while the Edge device holds a second portion. This segmentation allows processing to occur locally at the WTRU for immediate response, while the Edge device provides supplementary processing, thereby reducing network latency while maintaining video quality.
Solution Approach 2:
The patent introduces a spatial dimension to the processing architecture by distributing the neural network model across multiple physical locations (WTRU and Edge device). Instead of single centralized processing, the system uses geographic distribution of compute resources to simultaneously achieve low latency (through local WTRU processing) and high video quality (through Edge device processing).
2Measurement precision
If more neural network data is transmitted to improve prediction accuracy, then user action prediction improves, but data transmission time increases
Solution Approach 1:
The patent implements partial action by transmitting only the necessary portions of neural network data. The WTRU receives and processes a first portion of the DNN model data locally, which is sufficient for immediate predictions. The Edge device processes a second portion, eliminating the need to transmit the entire model. This partial transmission approach maintains prediction accuracy while minimizing data transmission time.
Solution Approach 2:
The Edge device acts as an intermediary between the centralized cloud and the WTRU. It stores and processes a portion of the neural network model, serving as a local cache that reduces the frequency and volume of data transmissions to the WTRU. This intermediary role maintains prediction accuracy by ensuring the WTRU has access to necessary model data without requiring constant full-model transmissions.
Data Source
AI summary
Methods and apparatuses for signaling enhancement to meet low-latency requirements of cloud gaming applications in wireless communication networks are provided. In an example, a method implemented by a wireless transmit/receive unit (WTRU) includes transmitting a first message including neural network data and information indicating a first type of neural network data, the neural network data were marshaled into one or more byte arrays before transmission; receiving a first acknowledgement message indicating a second type of neural network data that an Edge device has received; receiving a second message including marshaled data based on the transmitted neural network data and the information; and transmitting a second acknowledgement message indicating a third type of neural network data that the WTRU has received.