Research on Internet of Things Signals Acquisition and Processing based on Edge Computation

The proliferation of Internet of Things (IoT) devices has led to the generation of vast quantities of data, placing considerable strain on conventional cloud computing infrastructure. Edge computing has emerged as a viable solution to address this challenge. In this study, we propose an Internet of Things (IoT) signal processing framework that is combined with distributed edge nodes. The objective is to improve data transmission efficiency and real-time performance. Firstly, in consideration of the constrained resources of edge devices, we have devised lightweight preprocessing algorithms, including frequency domain filtering based on Fourier transform and a denoising algorithm based on wavelet transform, with the objective of reducing the noise and redundant information present in the original signal. Subsequently, the adaptive compressed sensing technology was employed to compress the preprocessed signal, which resulted in a notable reduction in the bandwidth requirement for data transmission. Subsequently, in order to fully utilise the computing capabilities of edge nodes, we introduce a multi-task learning collaborative computing model. The model permits the sharing of computing tasks between multiple edge nodes, facilitates data collaboration between edge nodes through federated learning, and reduces computing redundancy and communication overhead. The experimental phase employs the Google Edge TPU Dataset to evaluate the dataset pertinent to the Internet of Things. The results indicate that the proposed methodology exhibits a notable enhancement in data processing latency and transmission efficacy in comparison to the conventional approach.