A DISCRETE-OPTIMIZATION WITH HIERACHICAL CLASSIFIER FOR ATTACK PREDICTION

Abstract

Machine learning techniques are designed to act as a buffer against attack like jamming and cloning that are initiated over wireless networks. A pre-trained classifier is provided to the transmitter so that it may assess the channel's condition depending on the type of sensing it is doing and decide what needs to be transmitted next. All acknowledgements made among the nodes and the current channel state are gathered by the learning approach to construct a learning model that properly predicts the consecutive transmission limitation caused by network jamming. Here, the goal of an inventive anti-clone detection approach is to diminish the quantity of jamming and clones found throughout the network model relative to stochastic jamming methods. The transmitter analyzes the power restrictions over the sensor networks using the Discrete Particle Swarm Optimization (DPSO). In this case, by analyzing the incoming samples, a Hierarchical Prototype-based classifier is modelled to reduce the computing time required to gather the training dataset. By using this defence mechanism, the broadcaster hopes to forecast the false prediction rate (FPR) and create a more accurate model that will produce a classifier that is dependable. To increase throughput and lower prediction error, the transmitter systematically detects floating attack over the network model and implements the defence mechanism to trick the injected clone.