Embedded Smart Sensor Network Architecture Based on Edge Computing

This paper focuses on the integration of edge computing in smart sensor networks. In Internet of Things (IoT) applications, traditional cloud computing models cause data processing delays due to long-distance data transmission. This transmission delay significantly impairs real-time data processing and feedback capabilities, particularly in time-sensitive applications. Edge computing can transfer data to local devices for processing so that data that is affected by network congestion does not have to be transmitted long distances to the cloud for processing. It can greatly reduce the delay caused by network congestion and reduce energy consumption. In addition, it also provides certain guarantees for data security. Building upon these characteristics, edge computing offers several key advantages. First, it enables efficient local data processing for real-time operation. In some industrial applications such as smart grids and intelligent transportation systems, edge computing can operate highly autonomously. This avoids the delay caused by network fluctuations when transmitting to the cloud. This paper explores the integration of lightweight machine learning models, particularly TinyML, in edge computing applications. These models will improve the data processing capabilities of edge devices. Finally, the article also proposes further ideas for this new architecture to make up for the shortcomings of the architecture in task allocation and data processing.