“…In addition, the anti-jamming PHC-based algorithm of Reference [50] can be further developed to provide an anti-jamming mechanism for VANETs at the initial phase of the learning process, while achieving low-complexity RL solutions are necessary to improve PLS performance against smart jammers. Next, in GPS-spoofing scenarios [52], online learning can further improve the security provided by the NN, while unsupervised ML, capable of handling unlabeled data, can perform classification prior to processing, reducing the delay and increasing the accuracy. To further improve the performance of current ML-inspired detection methods, future work could be devoted to combining these methods with other conventional detection techniques, such as camera images and videos, radar echoes, and acoustic recordings, and to exploiting the advantages of each method.…”