Lightweight privacy protection outsourcing electroencephalogram signal feature extraction method

Abstract

The invention provides a lightweight privacy protection outsourcing electroencephalogram signal feature extraction method. According to the method, an edge server is allowed to perform feature extraction on electroencephalogram signal data uploaded by a user on the premise that plaintext electroencephalogram signal data cannot be acquired; meanwhile, in order to realize privacy protection of the electroencephalogram signals, a series of safe outsourcing computing protocols are designed, so that feature extraction of the encrypted electroencephalogram signals is realized; in addition, in order to improve the signal-to-noise ratio of the electroencephalogram signals, spatial filtering is conducted on the electroencephalogram signals through TRCA; finally, in order to reduce response delay between the user and the server, an edge computing technology is introduced. It is ensured that privacy information related to original electroencephalogram data is not leaked. According to the technical scheme, the feature extraction of the outsourcing electroencephalogram signal is realized while the privacy security of related information in the electroencephalogram signal is ensured, and the local calculation and communication cost of a user is reduced.

Description

technical field [0001] The invention belongs to the technical field of EEG signal processing, and in particular relates to a lightweight privacy protection outsourced EEG signal feature extraction method. Background technique [0002] Electroencephalography (EEG) is an electrophysiological monitoring method that uses electrophysiological indicators to record changes in the postsynaptic potential of a large number of neurons during brain activity. When EEG signals are captured, application developers can extract features from these signals and use them to train their own models that can be used for a variety of predictive and inference tasks, such as inferring personal emotions, religious beliefs, and forecasting seizures. [0003] At present, the mainstream EEG acquisition methods are generally divided into two types: invasive and noninvasive. Non-invasive acquisition methods have been widely used due to their safety and practicality. However, signals collected in this wa...

AI Technical Summary

Problems solved by technology

[0006] The present invention is mainly aimed at the problem that using edge computing to extract EEG signal features may lead to security threats, that is, the problem that private information in uploaded EEG signals may be leaked

Images