Enhancing Physical Layer Security of Random Caching in Large-Scale Multi-Antenna Heterogeneous Wireless Networks

Wanli Wen, Chenxi Liu, Yaru Fu, Tony Q. S. Quek, Fu-Chun Zheng, Shi Jin

Abstract

This paper introduces a secure random caching framework for large-scale multi-antenna heterogeneous networks (HetNets) designed to protect data from both passive eavesdroppers and active jammers. By optimizing file distribution across base stations and utilizing multi-antenna beamforming to inject "artificial noise," the framework demonstrates a significant boost in transmission reliability without compromising data confidentiality.

Introduction: The Challenge of Secure Caching

As mobile data traffic continues to surge, caching popular content at the network edge—specifically at base stations (BSs)—has become a critical strategy for reducing latency and backhaul congestion. However, traditional security measures often conflict with the goals of caching. Standard cryptographic encryption typically generates unique content for each user request, which prevents the reuse of cached data for multiple users, thereby undermining the efficiency of a shared cache.

Furthermore, wireless environments are inherently susceptible to "broadcast" attacks, where passive eavesdroppers intercept signals and active jammers intentionally degrade link quality. This research addresses these vulnerabilities by employing Physical Layer Security (PLS), leveraging the physical characteristics of wireless channels to ensure both data reliability and secrecy.

Methodology: Stochastic Geometry and Artificial Noise

The proposed framework models a multi-tier HetNet where multi-antenna BSs serve single-antenna users in the presence of single-antenna jammers and eavesdroppers. Key components of the methodology include:

  • Stochastic Geometry: The spatial distribution of BSs and jammers is modeled using Poisson Point Processes (PPPs). This approach enables the derivation of closed-form expressions for network-wide performance metrics, such as the Average Reliable Transmission Probability (RTP) and Average Confidential Transmission Probability (CTP).
  • Artificial Noise (AN) Injection: BSs utilize their multiple antennas to transmit the desired signal toward the legitimate user while simultaneously "spraying" artificial noise into the null space of the user's channel. This noise effectively confuses eavesdroppers while remaining invisible to the intended recipient.
  • Secure Random Caching: Instead of a deterministic "most popular" caching strategy, the system employs a probabilistic "random caching" scheme.
  • Optimization Framework: A non-convex optimization problem is formulated to maximize the RTP while satisfying constraints on CTP and storage capacity. The problem is solved using the Gradient Projection Method (GPM) and the Convex-Concave Procedure (CCP).

Key Insights: The Security-Reliability Trade-off

The research reveals several critical trade-offs in secure caching:

  1. The Caching Surface: Increasing the number of BSs that cache a file improves reliability by placing the content closer to the user. However, it also increases the "attack surface," providing more opportunities for eavesdroppers to intercept transmissions.
  2. Jammers as Accidental Allies: While jammers generally degrade reliability (RTP), they can paradoxically enhance secrecy (CTP). If the legitimate user is closer to the BS than the eavesdropper, the jammer's interference may degrade the eavesdropper's link more severely than the user's link.
  3. Sub-optimality of "Most Popular" Caching: Under strict secrecy requirements, caching only the most popular files is often sub-optimal. The optimal strategy may involve caching less popular content to minimize overall interference and intercept risk.

Performance Evaluation and Results

The theoretical derivations were validated through extensive Monte Carlo simulations. As shown in Figure 2, the analytical expressions for RTP closely match the simulation results across various target transmission rates, confirming the robustness of the mathematical model.

Validation of analytical RTP models against Monte Carlo simulations across different target transmission rates.

The system also demonstrates significant resilience against active jamming. Figure 1 illustrates how the RTP decreases as jammer transmit power and density increase. Despite this degradation, the optimized caching distribution allows the network to maintain the best possible balance between reliability and secrecy under adversarial conditions.

Impact of jammer transmit power (a) and jammer density (b) on the Reliable Transmission Probability (RTP) for various target rates.

The proposed optimized scheme significantly outperforms baseline strategies like "Most Popular Caching" and "Uniform Caching," particularly in high-secrecy scenarios where baseline methods often fail to meet confidentiality constraints.

Conclusion and Future Directions

This work provides a robust framework for integrating security into edge caching for large-scale networks. While effective, the model currently assumes single-antenna adversaries and perfect Channel State Information (CSI) for AN injection. Future research could explore scenarios involving multi-antenna colluding attackers and the impact of imperfect CSI on security performance.

Summary updated

BibTeX
@article{wen2020enhancing,
  author = {Wanli Wen and Chenxi Liu and Yaru Fu and Tony Q. S. Quek and Fu-Chun Zheng and Shi Jin},
  title = {{Enhancing Physical Layer Security of Random Caching in Large-Scale Multi-Antenna Heterogeneous Wireless Networks}},
  journal = {IEEE Trans. Inf. Forensics Secur.},
  year = {2020},
  doi = {10.1109/tifs.2020.2976961},
}
Figure