Cloud-Edge Distributed IoT Analytics for Bandwidth and Privacy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current cloud-based analytics face issues such as privacy invasion, network dependency, high bandwidth consumption, and slow response times due to the need for data transmission to a distant cloud, while edge computing struggles with processing large data and varies in performance across different edges, limiting its ability to handle complex computations and consistent model allocation.
Innovation Solution
A cloud-edge distributed operation method where the cloud creates multiple models and allocates them to edges based on their capabilities, allowing edges to perform lightweight operations and the cloud to handle in-depth learning and analytics, with data preprocessing and separation techniques to protect privacy and optimize bandwidth usage, and task modularization for distributed processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data is transmitted to cloud for analytics, then professional and advanced analytics can be performed, but network bandwidth is consumed and response time increases
Solution Approach 1:
The patent segments analytics tasks into two categories: simple field-level analytics performed at edge devices and complex professional analytics performed in the cloud. This segmentation allows critical analytics to be performed locally without consuming network bandwidth, while still enabling deep analysis for tasks that require cloud resources.
Solution Approach 2:
The patent implements preliminary data processing and filtering at the edge before transmission to the cloud. By performing initial analytics and preprocessing locally, the system reduces the volume of data that needs to be transmitted to the cloud, thereby reducing bandwidth consumption while maintaining analytical depth for transmitted data.
2Measurement precision
If data is transmitted to cloud for analytics, then advanced analysis can be performed, but privacy invasion may occur
Solution Approach 1:
The patent segments data processing between edge and cloud, keeping sensitive data processing at the edge where it remains local. Only anonymized or aggregated results are transmitted to the cloud, maintaining analytical depth while protecting privacy by preventing exposure of raw sensitive data.
Solution Approach 2:
The patent extracts and removes sensitive information from data before cloud transmission. By separating sensitive attributes from analytical data and processing them locally at the edge, the system enables cloud analytics on non-sensitive data while protecting privacy through extraction of harmful elements.
3Speed
If analytics are performed at edge, then response speed improves, but processing capability is limited compared to cloud
Solution Approach 1:
The patent segments analytics workloads by complexity, assigning time-critical simple analytics to edge devices for fast response and computationally intensive complex analytics to the cloud for high processing capability. This segmentation allows the system to achieve both fast response speeds and high processing capabilities for different task types.
4Reliability
If edge computing is used, then network dependency is reduced, but difficulty in allocating models consistently across different edges increases
Solution Approach 1:
The patent changes the parameter of model allocation from manual configuration to automated selection based on edge device characteristics. By dynamically adjusting model allocation decisions according to device capabilities, data types, and performance requirements, the system achieves consistent and optimal model distribution across heterogeneous edge devices without manual intervention.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Disclosed herein are a method and system for distributed operation between a cloud and an edge in an Internet of Things (IoT) computing environment including the edge and the cloud, the edge being connected to things to collect data from the things and transmit a control signal to the things, and the cloud being configured to receive data from the edge, process the data, and transmit a result of the processing to the edge. Traffic and load of the cloud for data processing are adaptively allotted to the edge depending on the situation such that the data is processed through cooperation between the edge and the cloud, and optimum processing results are provided. The distributed operation of the edge and cloud is performed so as to shorten the response time, to make the best of processing power, to be network-independent, and to protect data as much as possible.