TL;DR
This paper introduces a blockchain-based framework to improve resource sharing and access control for Virtual Network Functions (VNFs). It uses a game-theoretic model and a greedy algorithm to ensure fair profit distribution between VNF providers and infrastructure owners, while a novel blockchain-based access control mechanism enhances security and efficiency.
What Problem Does This Paper Address?
Network Function Virtualization (NFV) allows network functions (VNFs) to run on general-purpose servers, offering flexibility. However, deploying complex Service Function Chains (SFCs) involves multiple steps, and current VNF resource sharing often overlooks the financial interests of both VNF providers and infrastructure owners, leading to inefficient resource utilization and potential conflicts. Furthermore, securing access to these shared resources in a multi-tenant environment is a significant challenge, with traditional methods often relying on untrustworthy third parties.
Methods
The paper proposes a three-pronged approach:
- Resource Sharing Game Model: A game-theoretic model is developed to capture the strategic interactions between VNF instances (Resource Owners - ROs) and Infrastructure Providers (IPs). This model aims to optimize profits for both parties by considering factors like resource demand, cost, and a novel "relevance" parameter that quantifies potential conflicts between VNFs.
- Greedy Matching Algorithm: To operationalize the game model, a greedy matching algorithm is introduced. This algorithm efficiently matches VNF resource demands with available shared resources from ROs, considering their pricing strategies and the IP's profit maximization objective.
- Blockchain-Based Access Control: A novel access control mechanism is designed using blockchain technology and attribute-based encryption (ABE). This system securely stores cryptographic keys and access policies on the blockchain, enabling fine-grained control without relying on third-party authorities. It also incorporates an Adaptive Bloom Filter to obscure access policies, enhancing privacy.
Datasets
Not discussed in the paper. The evaluation uses simulations on the Open Network Automation Platform (ONAP) and Ethereum.
Key Insights and Ideas
- Incentivizing Resource Sharing: The core idea is that by modeling resource sharing as a game with clear profit incentives for both ROs and IPs, more efficient resource utilization can be achieved.
- The "Relevance" Parameter: Introducing a parameter that quantifies the potential for conflict between VNFs (e.g., timing or spatial overlap) is crucial for realistic profit modeling and strategic pricing.
- Blockchain for Trust and Transparency: Leveraging blockchain for storing access policies and keys eliminates the need for trusted third parties, enhancing security and providing an immutable audit trail for resource access.
- Greedy vs. Optimal: While a greedy approach might not always find the absolute optimal social utility, it offers a significant performance improvement and efficiency gain, making it practical for real-world deployment.
Key Results
- Performance Improvement: The proposed greedy matching algorithm achieves up to an 8.1 times performance improvement compared to the standard Stackelberg game solution, with only a marginal sacrifice in optimal social utility.
- Efficient Access Control: The blockchain-based access control mechanism demonstrates excellent encryption efficiency and moderate decryption efficiency, outperforming traditional CP-ABE methods in certain aspects.
- Scalability: The greedy algorithm shows remarkable runtime stability and minimal sensitivity to variations in resource demand and the number of ROs, indicating good scalability.
- Social Utility: The proposed system generally leads to increased social utility, especially when resource scarcity increases, incentivizing more resource sharing.
Limitations and Future Work
- Network Traffic Prediction: The paper acknowledges that network traffic prediction is not addressed. Integrating machine learning for traffic forecasting could further enhance pricing strategies for ROs.
- Correlation Parameter: The current model primarily uses a negative correlation parameter. Future work could explore more complex relationships and traffic sequencing among VNFs.
- Greedy vs. Optimal Trade-off: While the greedy algorithm is efficient, it doesn't always achieve the absolute optimal social utility. Future research could explore methods to bridge this gap.
Summary updated