Trust Anchors in Software Defined Networks

Paladi, Nicolae; Karlsson, Linus; Elbashir, Khalid

doi:10.1007/978-3-319-98989-1_24

Cited by 10 publications

(8 citation statements)

References 43 publications

(44 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in the unlikely case that a layer does not fit in available TEEs, the network design needs to be adjusted with smaller, but more layer(s), or a smaller training batch size. We also assume that there is a secure way to bootstrap trust between the server TEE and each of the client device TEE (e.g., using a slightly modified version of the SIGMA key exchange protocol [32,77], or attested TLS [30]), and that key management mechanisms exist to update and revoke keys when needed [55]. Finally, we assume that the centralized server will forward data to/from its TEE.…”

Section: Threat Model and Assumptionsmentioning

confidence: 99%

PPFL: Privacy-preserving Federated Learning with Trusted Execution Environments

Haddadi

Katevas

et al. 2021

Preprint

View full text Add to dashboard Cite

We propose and implement a Privacy-preserving Federated Learning (𝑃𝑃𝐹 𝐿) framework for mobile systems to limit privacy leakages in federated learning. Leveraging the widespread presence of Trusted Execution Environments (TEEs) in high-end and mobile devices, we utilize TEEs on clients for local training, and on servers for secure aggregation, so that model/gradient updates are hidden from adversaries. Challenged by the limited memory size of current TEEs, we leverage greedy layer-wise training to train each model's layer inside the trusted area until its convergence. The performance evaluation of our implementation shows that 𝑃𝑃𝐹 𝐿 can significantly improve privacy while incurring small system overheads at the client-side. In particular, 𝑃𝑃𝐹 𝐿 can successfully defend the trained model against data reconstruction, property inference, and membership inference attacks. Furthermore, it can achieve comparable model utility with fewer communication rounds (0.54×) and a similar amount of network traffic (1.002×) compared to the standard federated learning of a complete model. This is achieved while only introducing up to ∼15% CPU time, ∼18% memory usage, and ∼21% energy consumption overhead in 𝑃𝑃𝐹 𝐿's client-side. CCS CONCEPTS• Security and privacy → Privacy protections; Distributed systems security; • Computing methodologies → Distributed algorithms.

show abstract

Section: Threat Model and Assumptionsmentioning

confidence: 99%

PPFL: Privacy-preserving Federated Learning with Trusted Execution Environments

Haddadi

Katevas

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Section: Threat Model and Assumptionsmentioning

confidence: 99%

PPFL

Haddadi

Katevas

et al. 2021

Proceedings of the 19th Annual International Conference on Mobile Systems, Applications, and Services

123

View full text Add to dashboard Cite

We propose and implement a Privacy-preserving Federated Learning () framework for mobile systems to limit privacy leakages in federated learning. Leveraging the widespread presence of Trusted Execution Environments (TEEs) in high-end and mobile devices, we utilize TEEs on clients for local training, and on servers for secure aggregation, so that model/gradient updates are hidden from adversaries. Challenged by the limited memory size of current TEEs, we leverage greedy layer-wise training to train each model's layer inside the trusted area until its convergence. The performance evaluation of our implementation shows that can significantly improve privacy while incurring small system overheads at the client-side. In particular, can successfully defend the trained model against data reconstruction, property inference, and membership inference attacks. Furthermore, it can achieve comparable model utility with fewer communication rounds (0.54×) and a similar amount of network traffic (1.002×) compared to the standard federated learning of a complete model. This is achieved while only introducing up to ∼15% CPU time, ∼18% memory usage, and ∼21% energy consumption overhead in 's client-side. CCS CONCEPTS• Security and privacy → Privacy protections; Distributed systems security; • Computing methodologies → Distributed algorithms.

show abstract

“…TLSonSGX guarantees that OvS authorities retain communication with SDN controllers and the cryptographic material they use [24]. This methodology can be paired with OFTinSGX to provide broader security assurances for OpenFlow switches.…”

Section: Related Workmentioning

confidence: 99%

Secure Software Defined Networks Controller Storage using Intel Software Guard Extensions

Qasmaoui¹,

Maleh²,

Haqiq³

2020

IJACSA

View full text Add to dashboard Cite

The SDN controller is the core of the softwaredefined network (SDN), which provides important network operations that needs to be protected from all type of threats. Many researches have been focusing on different layers of security regarding the SDN controller such as Anti-DDOS system or enforcement of TLS connection between the controller and the Open-vswitches. One of the major security threats targeting any program is the environment execution itself (e.g. Operating system and the hardware itself). Intel's Software Guard Extension (SGX) offers a sloid layer of security applied to applications by creating a Trusted execution environment. SDN controller relay on a storage module to keep sensitive data such as Flow Rules, users' credentials and configuration files. Protecting this side of the SDN controller is a must in term of security. To date, no work has been conducted considering SDN controller storage security using Intel SGX. This paper introduces an SGX enabled SDN controller. The new controller ensures the integrity and the confidentiality in a trusted execution environment by leveraging a recent hardware technology called intel SGX. This technology provides a trusted and secure enclave. Enclaves are sealed and unsealed by intel SGX attestation mechanisms to protect the executed code and data inside live memory and disk from being altered by any unauthorized access. High privileged codes such as the OS itself is kept from altering data inside enclaves. We implemented the Intel SGX using the Floodlight SDN controller running a real enabled Intel SGX hardware. Our evaluation shows that the SGX enabled SDN controller introduces a slightly observable performance overhead to the floodlight controller compared to advantages in term of security.

show abstract

Trust Anchors in Software Defined Networks

Cited by 10 publications

References 43 publications

PPFL: Privacy-preserving Federated Learning with Trusted Execution Environments

PPFL: Privacy-preserving Federated Learning with Trusted Execution Environments

PPFL

Secure Software Defined Networks Controller Storage using Intel Software Guard Extensions

Contact Info

Product

Resources

About