SYNCHRONIZATION STRATEGIES FOR IOT AND EDGE INTELLIGENCE APPLICATIONS

Abstract

The sentence we provided talks about the challenges and importance of synchronization in edge-based IoT systems, particularly in the context of new Edge AI applications. It explains that simply synchronizing clocks is not enough and that a synchronization system needs to consider factors like movement, network disconnections, and other related problems. This research proposes three task-based strategies and two redundancy-based strategies to address the synchronization problem. The main idea is to group devices into synchronized teams and identify synchronization points using a theoretical solution. The method uses a hierarchical design with controllers at different levels and devices at the leaf level. A late notification protocol is used by the device cluster to select the best synchronization point from pre-calculated options for time-aligned task execution. The suggested "fast synchronizer" solution improves performance compared to existing alternatives. The research focuses on creating a quick synchronization method using simulations based on real-world data. The algorithms' performance will be evaluated and compared to existing solutions, considering fault tolerance to ensure reliability. The findings and outcomes of the research will be thoroughly discussed, and suggestions will be made based on the results. The research proposes a synchronization system that addresses these issues because clock synchronization alone is not sufficient for such systems. It describes three task-based solutions and two redundancy-based strategies to achieve synchronization in edge-based IoT. Additionally, it suggests a hierarchical design with controllers at different levels and devices at the leaf level to coordinate IoT operations involving drones and the Internet of Vehicles. The primary goal of the "rapid synchronizer" is to group devices into synchronized teams and determine synchronization locations using a theoretical solution. A late notification protocol is used to quickly select the best synchronization point for time-aligned task execution from pre-calculated options. The research demonstrates that the fast synchronizer outperforms other solutions and offers significant performance benefits. The study aims to address synchronization challenges in edge-based IoT systems, specifically in Edge AI applications, with the aim of improving convergence, accuracy, and speed in distributed training processes while considering fault tolerance. A rapid synchronization approach is being developed and evaluated for its performance.