Data high-quality compression method fusing dictionary training and observation matrix optimization

Abstract

The invention discloses a data high-quality compression method fusing dictionary training and observation matrix optimization, deep mining of correlation between a dictionary and an observation matrixis realized through joint optimization of a sparse dictionary and the observation matrix, and the problem of actual sensor data acquisition is mainly solved. The method comprises the steps that a two-dimensional dictionary training method based on K-SVD idea is constructed to realize sparse representation of space-time data in IIoT; column coherence of a sensing matrix is minimized by optimizingcorresponding observation matrixes in space and time domains, so that a trained sparse dictionary is matched; and finally, a joint optimization method is provided, and trade-off consideration is performed between dictionary training and observation matrix optimization, so that the reconstruction error reaches a theoretical minimum value. Compared with an existing data compression scheme, the HQDCmethod provided by the invention can efficiently work in a real IIoT scene; meanwhile, the HQDC method is remarkably superior to other data compression schemes in reconstruction precision.

Description

technical field [0001] The invention belongs to the field of wireless sensor networks, and in particular relates to a high-quality data compression method based on fusion dictionary training and observation optimization. Background technique [0002] IoT has been extensively researched and applied in fields such as military reconnaissance, environmental monitoring, security systems, and industrial automation. In harsh and catastrophic environments, the unreliability of sensors makes it inefficient or even ineffective to deploy centralized static sink nodes to collect data from distributed sensors. In response to this problem, scholars have proposed many solutions such as distributed data storage (Distributed Data Storage, DDS), through the introduction of network redundancy, to achieve reliable data collection and robust network operation. A study of the prior art identified a hybrid CS cluster architecture. The architecture collects data via shortest path routing within t...

AI Technical Summary

Problems solved by technology

However, in the case of sufficient data, neither of their methods takes advantage of the structure in the data

Images