Fog Computing Terminal Incremental Learning for IoT Bandwidth Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing IoT systems face challenges in achieving accurate and user-specific prediction results due to the centralized computation approach, which requires high bandwidth and is inefficient in processing massive data from various connected devices.
Innovation Solution
A monitoring method utilizing a fog computing terminal that downloads an application packet from a cloud server, sets IoT devices into groups, receives sensor data, and executes incremental learning to obtain an incrementally learning model, allowing for localized prediction results closer to user needs, while employing blockchain technology for secure and efficient data management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If centralized computation is used to process massive IoT data, then data processing capability is improved, but bandwidth requirements increase and system efficiency deteriorates
Solution Approach 1:
The patent segments the centralized cloud computing system into a distributed architecture comprising cloud servers, fog computing terminals, and IoT devices. The fog computing terminal acts as an intermediate node that performs local data processing and filtering, segmenting the data processing workload from the centralized cloud server. This segmentation reduces the volume of data transmitted over the network, thereby reducing bandwidth requirements while maintaining data processing capability.
Solution Approach 2:
The patent introduces a new dimension of computation by deploying fog computing terminals at the network edge, creating a multi-layered computing architecture (cloud-fog-device). This dimensional change allows computation to occur closer to the data source, reducing the need for high-bandwidth communication between IoT devices and the cloud server while maintaining processing power.
2Power
If centralized computation is used to process massive IoT data, then data processing capability is improved, but system efficiency deteriorates
Solution Approach 1:
The system segments processing tasks between fog computing terminals and cloud servers. The fog terminal performs immediate local processing of sensor data, filtering and pre-processing information before transmission to the cloud. This segmentation enables parallel processing across multiple nodes, improving overall system efficiency while maintaining data processing capability.
Solution Approach 2:
The fog computing terminal performs preliminary data processing, filtering, and aggregation at the network edge before data is transmitted to the cloud server. This preliminary action reduces the volume of data requiring centralized processing and enables faster local responses, thereby improving system efficiency without sacrificing overall processing capability.
3Measurement precision
If incremental learning is executed at the fog computing terminal, then user-specific prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The system employs transfer learning where a general model is pre-trained on the cloud server using large-scale data, and then this pre-trained model is transferred to the fog computing terminal. The terminal performs incremental learning by fine-tuning the pre-trained model with local user-specific data. This preliminary action of pre-training reduces the computational burden and complexity at the edge device while maintaining the ability to achieve high user-specific prediction accuracy.
Solution Approach 2:
The fog computing terminal acts as an intermediary between the cloud server and the user's IoT devices. It hosts the incremental learning model that bridges general knowledge from the cloud and specific user data from local devices. This intermediary role allows the terminal to manage the complexity of learning algorithms while providing simplified, accurate predictions to end users.
Data Source
AI summary
A monitoring method based on Internet of things (IoT), a fog computing terminal and an Internet of things system are provided. The fog computing terminal downloads an application package from a cloud server, downloads a general model from the cloud server through the application package, sets one or more IoT devices to a device group to receive a plurality of sensing data from the device group, and executes an incremental learning based on the sensing data and the general model to obtain an incrementally learning model for controlling the device group by the incrementally learning model.