Privacy-Preserving Network Analytics and Virtual Network Functions in SDN and Cloud Environments Using Homomorphic Encryption

Privacy-Preserving Network Analytics and Virtual Network Functions in SDN and Cloud Environments Using Homomorphic Encryption

Yeganeh Behmadi1 Fatemeh Golzari2 Toktam Balforush3 Sepehr Fakhr Eskandari4 Behzad Molvai5 Mostafa Farhadi-Moghadam6

1) Computer Science Student, Department of Computer, Vahdat-Torbat Jam Institute of Higher Education, Email:
2) Computer Science Student, Department of Computer, Vahdat-Torbat Jam Institute of Higher Education, Email:
3) Computer Science Student, Department of Computer, Vahdat-Torbat Jam Institute of Higher Education, Email:
4) Department of Computer, Vahdat-Torbat Jam Institute of Higher Education, Email:
5) Corresponding Author, Department of Computer Science, Vahdat-Torbat Jam Institute of Higher Education, Email:
6) Department of Computer Science, Vahdat-Torbat Jam Institute of Higher Education, Email:

IOI Code : AAHD-ABHHKD
Publication : National Conference on Information Technology, Nanotechnology, Artificial Intelligence and Technological Futures Studies(itnaf.ir)
Abstract :
    The rapid adoption of Software-Defined Networking (SDN) and Network Function Virtualization (NFV) in multi-tenant cloud and 5G/6G environments has improved programmability and service flexibility, while simultaneously intensifying privacy risks. Centralized controllers and cloud-hosted Virtual Network Functions (VNFs) routinely process sensitive flow statistics and metadata under an honest-but-curious threat model. At the same time, end-to-end encryption, especially TLS 1.3, and regulations such as GDPR have further constrained traditional traffic inspection techniques. Existing privacy-preserving approaches, including differential privacy, trusted execution environments, and secure multi-party computation, either reduce analytical utility, introduce hardware-specific trust assumptions, or impose prohibitive communication overhead. This paper presents a unified framework based on the CKKS fully homomorphic encryption scheme to enable both privacy-preserving network analytics and core virtual network functions without decrypting data at intermediate entities. The proposed three-layer architecture consists of the Data Plane, the SDN Control Plane, and the Cloud Analytics/VNF Platform. It supports encrypted flow-statistics aggregation, traffic classification, anomaly detection, load-balancing decisions, policy and firewall evaluation, and multi-tenant VNF orchestration. The framework incorporates optimized CKKS encoding, ciphertext packing, noise management, and integration with existing SDN northbound interfaces. Security analysis confirms semantic security under the Ring Learning With Errors assumption against honest-but-curious adversaries. Although computational overhead remains a practical limitation for ultra-high-speed network environments, the framework demonstrates that strong confidentiality can be achieved with acceptable analytical accuracy and functional correctness. Therefore, it offers a viable path toward privacy-aware SDN and cloud deployments.

Keywords : homomorphic encryption ckks software defined networking network function virtualization privacy preserving analytics virtual network functions