A Semantic Communication Method for Intelligent Internet of Things Based on Images

For image processing tasks in intelligent Internet of Things (IoT) devices, the traditional method is to upload the raw image data of IoT devices to an edge/cloud server for processing, and then return the processed results to the IoT end. However, this method may cause significant delay problems and affect the real-time performance of IoT devices and application effects for large data tasks. To solve this problem, this paper proposes an image semantic communication method that reduces the amount of data transmitted by using a semantic encoder to extract key category information and spatial arrangement information from the original image. At the receiving end, key information is regenerated through semantic deconstruction and reconstruction methods using a decoder, so that the receiving end can use the generated image data to perform tasks normally. Simulation results show that this method outperforms traditional communication methods in terms of improving latency performance.