Security analysis and prediction of multi-relay networks over Fisher-Snedecor F fading channels

This paper investigates the physical layer security (PLS) of multi-relay networks over Fisher-Snedecor F fading channels. Precise expressions for outage probability (OP), intercept probability (IP), secure outage probability (SOP), and probability of strictly positive secrecy capacity (SPSC) are derived based on this considered model, which the SOP and SPSC are provided in a unified manner using Meijer's G-function. Moreover, simulation consequences of Monte Carlo experiments validate derived mathematical expressions. Through theoretical analysis and simulation results, the interesting results are that larger m sRD , smaller m RD , mu RE and m sRE the better secrecy performance of this system. Because deep learning has characteristics of low latency, strong learning ability, and adaptability, in order to pursue lower latency and universal applicability, we combine deep learning and wireless communication. On the basis of constructing a data set with exact closed-form expressions, the secrecy performance is predicted by utilizing a multi-layer perceptron (MLP) deep learning model. Moreover, by comparing the three algorithms of convolutional neural network (CNN), deep neural network (DNN), and two-layer long short-term memory network (LSTM), the MLP has the advantages of higher performance and lower complexity. Experimentations exhibit that the presented algorithm enhances the precision by 75.69% with LSTM for comparison and reduces the time complexity by 80.05% compared with DNN.