Repository of DUTH

Reputation-based solutions for trustworthy next-generation networks (Doctoral thesis)

Καπετανίδου, Ιωάννα Αγγελική/ Kapetanidou, Ioanna Angeliki

In the modern Internet era, user- and data-centricity have evolved into two fundamental pillars. This is mainly due to the vast expansion of data sources, particularly with the emergence of the Internet of Things (IoT). This necessitates specialized solutions to efficiently manage the ever-growing data traffic and support complex data types, such as multimedia content, while ensuring data reliability, attack resilience, and a high Quality of Experience (QoE) for end-users. In this dissertation, we leverage reputation-based trust as a key tool to address such challenges in two prominent next-generation networking paradigms: Information-Centric Networking (ICN) and Cloud Networking. In the first part of this dissertation, we explore how reputation-based trust can consolidate Information-Centric Networking; focusing on Named Data Networking (NDN), the de-facto ICN architecture, we explore reputation-based trust as a means to tackle some of its inherent vulnerabilities, particularly in its security and caching functionalities. In this context, our work is concerned with various use cases and network settings, aiming to showcase the potential of reputation in a wide range of application domains. We first demonstrate how reputation can be used as a complementary mechanism to existing security schemes in NDN ad hoc networks. We employ a reputation-based mechanism and evidently show its benefits in terms of usability and communication and computational costs. Secondly, we propose two distinct reputation-based mechanisms, each one designed to mitigate a different crucial NDN security threat, namely content poisoning and bitrate oscillation attacks. Our evaluation results indicate that reputation can be an effective countermeasure to achieve attack mitigation in both cases. More specifically, the first mechanism restricts significantly the spread of poisonous content in the network by avoiding notorious namespaces, while the second by adapting the caching policy based on the measured bitrate oscillations during multimedia streaming sessions, exhibits QoE performance close to the respective in no-attack scenarios. Last but not least, we introduce a reputation-based caching policy, specially tailored to ensuring fault- and energy-awareness in NDN-based IoT networks. The evaluation results revealed its benefits in terms of improved cache utilization and energy efficiency. The second issue addressed in this dissertation is QoE-aware cloud service monitoring, emphasizing the need for integrating exclusively trustworthy user feedback in the monitoring plane. We introduce Feed4Cloud, a blockchain-based solution that uses reputation to ensure fair QoE-based service evaluation. The service reputation score is integrated into the monitoring plane to, ultimately, enable improved service configuration and advanced cloud-to-edge orchestration. Corresponding evaluation results demonstrate that Feed4Cloud is able to correctly identify and filter out invalid user feedback, even if the attack tactic is quite sophisticated, while QoE-based service evaluation is consistently conducted in a trustworthy manner. Implementation-wise, in the first dissertation part, the proposed mechanisms have been developed and validated in simulation environments, while, in the second part, we provide a real-world proof-of-concept prototype. Through extensive evaluations, this dissertation demonstrates the effectiveness of reputation-based trust mechanisms in enhancing security and maintaining high user QoE in both ICN and cloud networks. Overall, this dissertation provides insights into the practical implementation and performance benefits of reputation-based approaches, contributing to the development of robust and adaptive trust frameworks for next-generation networks.