Abstract
The flexible, interoperable, and innovative network architecture of Open Radio Access Network</p><p>(Open RAN) offers numerous advantages but also presents new security challenges that require</p><p>sophisticated defense mechanisms. This thesis addresses the pressing security issues associated</p><p>with Open RAN and introduces a series of groundbreaking solutions designed to strengthen</p><p>the resilience of contemporary telecommunications infrastructures. Through systematic research</p><p>and experimentation, this work explores a range of security strategies, beginning with</p><p>conventional Intrusion Detection Systems (IDS) and progressing to the use of blockchain and</p><p>Generative Artificial Intelligence (GenAI) for adaptive and more intelligent threat detection.</p><p>This research study starts with an exhaustive review of the current literature on Open RAN</p><p>security, pinpointing critical vulnerabilities and detailing the current threat landscape. This</p><p>analysis not only identifies substantial gaps in existing security frameworks but also categorizes</p><p>the specific threats pertinent to Open RAN environments, thereby laying the groundwork for</p><p>the development of specialized security solutions for Open RAN.</p><p>Building on this foundation, the thesis introduces an innovative near-real-time IDS custom-built</p><p>for Open RAN ecosystem, engineered to mitigate a variety of network threats. By employing</p><p>Stack Ensemble Learning (SEL) and dimensionality reduction technique, the IDS is optimized</p><p>for integration within the Near-Real-Time Radio Intelligent Controller (Near-RT RIC). The key</p><p>advancement of this IDS is the SEL classifier, which synergizes multiple learning algorithms to</p><p>enhance threat detection. Evaluations based on metric such as accuracy, recall, precision, and</p><p>F1-score show that the system enhances security without compromising network performance.</p><p>The study delves into the synergy between blockchain and artificial intelligence/machine learning</p><p>(AI/ML) to strengthen Open RAN security. By merging the robust security properties of</p><p>blockchain with the adaptive and predictive capabilities of AI/ML, the thesis conducts a thorough</p><p>experiment on intelligent security measures using synthetic and real-world Open RAN</p><p>datasets. The findings show that this combined approach significantly strengthens the defense</p><p>of Open RAN networks by enabling the immutable recording of events, including threat detection,</p><p>ensuring data integrity, resilience against emerging threats, and supporting a sustainable</p><p>digital infrastructure for 5G and beyond.</p><p>Furthermore, this thesis explores a novel application of GenAI in intrusion detection for Open</p><p>RAN. GenAI models simulate complex attack scenarios and adaptively learn from them, empowering</p><p>Open RAN networks to adapt, predict and neutralize potential threats more effectively.</p><p>In sum, this thesis concludes by discussing the limitations of the research and suggesting future</p><p>directions to further advance the security of Open RAN networks.</p>