Deep Learning Aided Physical-Layer Security: The Security Versus Reliability Trade-Off

This paper considers a communication system whose source can learn from channel-related data, thereby making suitable choice of parameters for security improvement. The the is optimized using deep neural networks (DNNs). More explicitly, associated vs reliability trade-off problem characterized in terms symbol error probabilities and discrete-input continuous-output memoryless channel (DCMC) capacities. A pair loss functions were defined by relying on Lagrangian monotonic-function based techniques. These then used managing learning/training process DNNs finding near-optimal solutions to non-convex problem. technique was shown approach performance exhaustive search. We concluded characterizing intercept probability outage probability.