Edge Neural Network Partitioning for Autonomous Vehicle Data Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face significant data traffic and battery consumption issues when sending large sensor data to remote cloud servers for processing, degrading network performance and consuming excessive power.
Innovation Solution
Implementing edge processing by intelligently partitioning the computation of artificial neural networks (ANNs) across multiple devices, including cloud, edge servers, and vehicles, where less data is sent over networks by processing certain layers closer to the data source, reducing data traffic and battery usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor data is sent to remote cloud servers for processing, then processing capability is improved, but data traffic increases and battery power is consumed excessively
Solution Approach 1:
The patent segments the neural network processing into two parts: a compressed version of the ANN is deployed at the edge device (vehicle) for local processing, while the full ANN remains at the cloud server. This segmentation allows critical processing to occur locally with minimal energy consumption, while maintaining the option to use full cloud processing when needed.
Solution Approach 2:
The patent introduces an intermediary compressed neural network model that acts as a mediator between the edge device and the full cloud-based ANN. This compressed model processes data locally at the edge, reducing the need to transmit large amounts of sensor data to the cloud, thereby reducing both data traffic and energy consumption while still providing accurate processing results.
2Reliability
If sensor data is sent to remote cloud servers for processing, then processing capability is improved, but data traffic increases and network performance degrades
Solution Approach 1:
The patent segments the neural network processing into two parts: a compressed version of the ANN is deployed at the edge device (vehicle) for local processing, while the full ANN remains at the cloud server. This segmentation allows critical processing to occur locally with minimal energy consumption, while maintaining the option to use full cloud processing when needed.
Solution Approach 2:
The patent introduces an intermediary compressed neural network model that acts as a mediator between the edge device and the full cloud-based ANN. This compressed model processes data locally at the edge, reducing the need to transmit large amounts of sensor data to the cloud, thereby reducing both data traffic and energy consumption while still providing accurate processing results.
3Quantity of substance
If edge processing is implemented by partitioning neural network computation, then data traffic is reduced, but device complexity increases
Solution Approach 1:
The patent applies parameter changes by compressing the neural network model to create a smaller version that can be deployed at the edge device. This compression involves changing the parameters of the neural network (reducing the number of weights and connections) while maintaining its essential processing capabilities. The compressed model strikes a balance between model size for edge deployment and processing accuracy.
4Measurement precision
If all sensor data is processed at the cloud server, then processing accuracy is maintained, but energy consumption and data traffic increase
Solution Approach 1:
The patent segments the neural network processing into two parts: a compressed version of the ANN is deployed at the edge device (vehicle) for local processing, while the full ANN remains at the cloud server. This segmentation allows critical processing to occur locally with minimal energy consumption, while maintaining the option to use full cloud processing when needed.
Solution Approach 2:
The patent introduces an intermediary compressed neural network model that acts as a mediator between the edge device and the full cloud-based ANN. This compressed model processes data locally at the edge, reducing the need to transmit large amounts of sensor data to the cloud, thereby reducing both data traffic and energy consumption while still providing accurate processing results.
Data Source
AI summary
Methods, systems, and apparatuses related to edge processing of sensor data using a neural network to reduce network traffic to and/or from a server. In one approach, a cloud server processes sensor data from a vehicle using an artificial neural network (ANN). The ANN has several layers. Based on analyzing at least one characteristic of the sensor data received from the vehicle and/or a context associated with processing the sensor data, the cloud server determines to send one or more of the layers of the ANN for edge processing on the vehicle itself. In other cases, the cloud server decides to send the one or more layers to an edge server device located on a communication path between the vehicle and the cloud server. The edge processing reduces network data traffic.