Homomorphic Encryption-Enabled Federated Learning for Privacy-Preserving Intrusion Detection in Resource-Constrained IoV Networks (2407.18503v1)

Abstract: This paper aims to propose a novel framework to address the data privacy issue for Federated Learning (FL)-based Intrusion Detection Systems (IDSs) in Internet-of-Vehicles(IoVs) with limited computational resources. In particular, in conventional FL systems, it is usually assumed that the computing nodes have sufficient computational resources to process the training tasks. However, in practical IoV systems, vehicles usually have limited computational resources to process intensive training tasks, compromising the effectiveness of deploying FL in IDSs. While offloading data from vehicles to the cloud can mitigate this issue, it introduces significant privacy concerns for vehicle users (VUs). To resolve this issue, we first propose a highly-effective framework using homomorphic encryption to secure data that requires offloading to a centralized server for processing. Furthermore, we develop an effective training algorithm tailored to handle the challenges of FL-based systems with encrypted data. This algorithm allows the centralized server to directly compute on quantum-secure encrypted ciphertexts without needing decryption. This approach not only safeguards data privacy during the offloading process from VUs to the centralized server but also enhances the efficiency of utilizing FL for IDSs in IoV systems. Our simulation results show that our proposed approach can achieve a performance that is as close to that of the solution without encryption, with a gap of less than 0.8%.

• The paper presents a novel FL framework that leverages homomorphic encryption to process encrypted data directly, ensuring privacy during intrusion detection in IoV networks.
• The framework mitigates resource constraints by offloading computational tasks to centralized servers while maintaining a performance gap of less than 0.8% compared to non-encrypted methods.
• The proposed CKKS-based training algorithm, utilizing SIMD and bootstrapping, achieves approximately 91% accuracy on the Edge-IIoT dataset, demonstrating both efficiency and security.

Homomorphic Encryption-Enabled Federated Learning for Privacy-Preserving Intrusion Detection in Resource-Constrained IoV Networks

The paper proposes a framework to address data privacy challenges and computational resource limitations in Federated Learning (FL)-based Intrusion Detection Systems (IDSs) for Internet-of-Vehicles (IoV) networks. With the increasing convergence of vehicular ad hoc networks (VANETs) and Internet of Things (IoT) sensors in IoVs, the need for robust cybersecurity solutions is paramount. The proposed framework leverages homomorphic encryption (HE) to enhance data privacy while offloading computationally intensive tasks to centralized servers.

Key Contributions:

Novel Framework for Privacy Preservation:

The primary contribution is the development of a privacy-preserving FL framework where encrypted data is processed directly without decryption, utilizing quantum-secure encryption techniques. This ensures that Vehicle Users’ (VUs) data remains confidential throughout the transmission and processing stages.

Handling Computational Constraints:

Given the limited computational resources of VUs, the framework allows for the offloading of encrypted data to a centralized server. This approach mitigates local resource limitations but introduces a significant challenge—the need to perform computations on encrypted data, which the framework successfully addresses using advanced HE techniques.

Advanced Training Algorithm:

The paper presents a training algorithm tailored for encrypted data using the Cheon-Kim-Kim-Song (CKKS) scheme. This algorithm incorporates Single Instruction Multiple Data (SIMD) and bootstrapping to perform operations on encrypted data efficiently. Such a design maintains data privacy without considerably compromising the learning performance.

Performance Evaluation:

Simulation Setup:

The framework's efficacy was validated using the Edge-IIoT dataset, consisting of 31,400 samples across six classes representing normal traffic and various attack types. The dataset was preprocessed and utilized to train a fully connected neural network with two hidden layers.

Evaluation Metrics:

Performance was evaluated using accuracy, precision, and recall metrics derived from the confusion matrix. The results showed that while the proposed framework achieved slightly lower performance metrics than non-encrypted FL frameworks, the gap was less than 0.8%.

Results:

• Accuracy: Approximately 91% for both 2-VU and 3-VU scenarios when offloading 10% and 20% of local data.
• Precision and Recall: Consistently high, with only marginal declines observed compared to frameworks that did not employ encryption.

Theoretical and Practical Implications:

Enhancing IoV Security:

This research significantly impacts the practical deployment of IDSs in IoV networks by demonstrating that HE can be effectively integrated with FL to uphold data privacy without major performance degradation. This ensures that VUs' privacy is preserved even when data is offloaded to centralized servers for processing.

Future Developments in AI:

The framework sets a precedent for future developments in AI where privacy-preserving methodologies are essential. The integration of HE with machine learning models is likely to see broader applications in other domains requiring stringent data privacy measures, such as healthcare and finance.

In conclusion, this paper provides a substantial advancement in the field of IoV cybersecurity through a privacy-preserving FL framework. The framework's successful integration of homomorphic encryption with federated learning addresses both computational constraints and privacy concerns, paving the way for robust, privacy-respecting IDS solutions in resource-constrained environments. Future research may focus on optimizing the HE techniques to further reduce the performance gap and explore additional applications of these privacy-preserving methods in other sectors.