Deep Learning based approach for detection of Resource Depletion Attacks in 5G Networks
Mansi Sharma, M Purna Rama Satya Sai, Sriram Sankaran · 2024
Resource Depletion Attacks pose significant challenges to the stability and efficiency of 5G networks, with potential repercussions on service availability, network performance, and user experience. This problem is motivated by the need to understand the complexities of RDAs and their specific impact on the 5G Core networks, to develop effective countermeasures. First, we investigate the vulnerabilities of 5G networks, which make them susceptible to Resource Depletion Attacks. By identifying and analyzing these weaknesses, we aim to gain a comprehensive understanding of potential attack vectors. The problem of modelling and detecting Resource Depletion Attacks (RDA) on 5G Core, aiming to enhance their security and resilience against cyber threats is discussed in this paper. Our proposed attack model will incorporate essential 5G network elements, including base stations, user equipment, and core network components while integrating realistic traffic patterns and usage behaviours. Finally, a Deep learning-based mechanism will be evaluated to counter RDA on 5G Core. We incorporated two deep learning algorithms, specifically LSTM and GRU, which exhibited robust performance by achieving accuracies of $97.62 \%$ and $99.07 \%$, respectively. The outcomes of our work will contribute to the development of resilient 5 G networks capable of withstanding sophisticated cyber threats, ensuring uninterrupted and secure communication and connectivity services by addressing the challenges posed by Resource Depletion Attacks.