Computationally Efficient Approach for 6G-AI-IoT Network Slicing and Error-Free Transmission

Abstract

Many smart gadgets are connecting to the Internet, and Internet-of-Things (IoT) technologies are enabling a variety of applications. Artificial intelligence (AI) of Things (AIoT) devices are anticipated to possess human-like decision-making, reasoning, perception, and other capacities with the combination of AI and IoT. AIoT gadgets are expected to be extensively utilized across several domains, as anticipated by 6G networks. With AI's steady advancements in speech recognition, computer vision, and natural language processing—not to mention its ability to analyze large amounts of data—semantic communication is now feasible. A new paradigm in wireless communication is opened by semantic communication, which seeks to explore the meaning behind the bits and only transmits the information that may be used, as opposed to attaining error-free transmission. The combination of IoT with AI provides prominent features to overcome various important issues in cloud computing networks. However, there is bottleneck of delay and precision. Therefore, this paper proposed a new method to overcome this problem. First, the network slicing feature maps were extracted by convolutional neural networks. Next, the processing delay is reduced by semantic compression. Simulation results show that the proposed approach makes 99.2% reduction in communication complexity and an 80% reduction in transmission delay as compared with traditional methods. Taking the Resnet18 network as an example, the running time of the semantic communication method is only 0.8% of the traditional method.