Learning-Based Efficient Sparse Sensing and Recovery for Privacy-Aware IoMT

Due to the inherent openness of wireless channels and the restriction of communication resources and energy supply, the privacy protection of the sensing data transmission in the security-critical Internet of Medical Things (IoMT) has become a great challenge. In order to guarantee the privacy of IoMT sensing and transmission in a wireless wiretap channel and reduce the power consumption, a privacy-aware sensing and transmission scheme with the name of sparse-learning-based encryption and recovery (SLER) is proposed. The sparse sensing signal is compressed and encrypted at the IoMT devices in the encryption stage and transmitted to the network coordinator or edge devices, where the sparse signal is accurately recovered via sparse learning in the decryption stage. The encryption stage is conducted based on compressed sensing. The decryption stage utilizes a model-based sparsity-aware deep neural network to accurately recover the sensing signal, whose sparse features are extracted to decrease the required size of measurement signals and increase the spectrum efficiency. The secrecy performance of the proposed SLER algorithm is theoretically analyzed. Experiments of electrocardiogram (ECG) signal transmission are performed as a typical IoMT application. The experimental results show that the proposed scheme can effectively guarantee the transmission secrecy against eavesdropping, while improving the spectrum efficiency and energy efficiency compared to other existing methods.