A Study on Sophisticated Web Phishing Attacks and Their Mitigation Techniques

Abstract

Phishing is a tactical technique practiced by cybercriminals, wherein the target sys- newlinetems are approached, made vulnerable, and exploited. A phisher who engages in phish- newlineing is always creative, calculative, and persistent. This potentially leads to an increase in the success rate of phishing, and even individuals with technical expertise can fall victim to phishing campaigns. Objectives of the thesis to study the existing state-of- newlinethe-art adversarial phishing techniques, to develop more robust solutions for defending newlinethe phishing campaigns and to test and validate the proposed solutions in real time. newlineThere are three research contributions are provided for achieving the objectives. The newlinefirst work focuses on the effective detection of malicious URLs, utilizing lexical pa- newlinerameters for feature extraction, employing deep learning models (LSTM and GRU) newlinewith TensorFlow, and optimizing for faster inference on edge devices. Unlike previ- newlineous studies, which looked at online content, URLs, and traffic numbers, this work aims newlineto focus only on the text in the URL which makes it quicker, and thereby zero-day newlineassaults could be caught at the earliest. RNN has been optimized so that it might be newlineutilized on tiny devices like mobiles, and raspberry pi without sacrificing the inference time. The second work focuses on the Bat Algorithm, a swarm intelligence approach inspired by micro-bat behavior, incorporating parameters and mimicking echolocation for solution optimization to increase phishing detection efficiency. The results of utilizing deep neural networks heavily depend on the setting of different learning parameters. newlineIn this work, bat algorithm based approach to parameter setting of deep learning neu- newlineral network. The proposed approach to the classification of phishing websites is able newlineto improve the detection when compared to existing algorithms. The third work intro- newlineduces a novel approach, integrating the Firefly Optimization Algorithm with advanced newlinefeature engineering for phishing detection, achieving increased accu