Physical‐Layer Location Verification by Machine Learning

This chapter analyzes machine learning (ML) algorithms for in region location verification (IRLV) with emphasis on multiple‐layer neural network (NN) and support vector machine (SVM) approaches. As these solutions efficiently match the performance of the Neyman‐Pearson (NP) hypothesis test at convergence, they turn out to be both effective and efficient in terms of computational complexity, resource consumption, and processing latency. The chapter first describes the IRLV system model and revises the optimal hypothesis testing framework. It then presents the ML IRLV procedure and quickly reviews the fundamentals of NNs and SVMs. The chapter also discusses the optimality of ML‐based classification, and reviews theoretical results on the complexity‐performance trade‐off of ML solutions, which will be used to assess performance of methods based on NNs. It further presents the results obtained over a dataset of experimental data with attenuation values of a cellular system.