The code.m serves as the main body of the code, encompassing the primary workflow of the experiment. syncsignal.m is responsible for generating the synchronization signal, while denoisesignal.m and sec_channel.m handle noise reduction. The noise reduction process involves implementing denoising techniques. matchfilter.m is utilized for matched filtering. The function get_range_and_energy.m is employed to acquire reference distance and energy information, essential for calculating ECF (Energy Compaction Factor). range_compensation.m and com_energy.m are employed to compensate for distance and energy, respectively. Lastly, feature_extraction.m is responsible for implementing the feature extraction functionality. model.kt is the Kotlin-based neural network design portion.
Ref: Zhaopeng Xu, Tong Liu, Ruobing Jiang, Pengfei Hu, Zhongwen Guo, Chao Liu*. AFace: Range-flexible Anti-spoofing Face Authentication via Smartphone Acoustic Sensing. UbiComp 2024. (CCF A)