Power Allocation for Energy Efficiency and Secrecy of Wireless Interference Networks

Zhichao Sheng, Hoang Duong Tuan, Ali Arshad Nasir, Trung Quang Duong, H. Vincent PoorIEEE Trans. Wireless Communications 2018 · DOI

Abstract

This paper presents efficient power allocation algorithms designed to maximize "worst-case" secrecy throughput and secure energy efficiency (SEE) in wireless interference networks. Addressing the practical challenge of imperfect channel state information (CSI), the authors develop robust, low-complexity path-following procedures. These algorithms ensure secure communication under various CSI scenarios, including perfect, partial, and robust models, by iteratively solving convex approximations of non-convex optimization problems.

Introduction

Wireless communication is inherently vulnerable to unauthorized interception due to its broadcast nature. Physical Layer Security (PLS) has emerged as a promising paradigm to safeguard data by exploiting the physical properties of the communication channel. However, implementing PLS in multi-user networks faces two significant challenges: managing mutual interference between users and maintaining high energy efficiency.

System model of a wireless interference network with an eavesdropper.

The network model considered involves multiple transmitter-user pairs operating in the presence of an eavesdropper. As shown in the system diagram above, each transmitter aims to send data to its intended user (UE) while contending with interference from other transmitters and the risk of wiretapping. Most existing research assumes perfect Channel State Information (CSI), which is often unattainable in real-world deployments. This work addresses this gap by optimizing power allocation under perfect, partial (statistical), and robust (uncertain) CSI scenarios.

Methods

The optimization of secrecy throughput and SEE is a non-convex problem, typically difficult to solve efficiently. While Difference of Convex (D.C.) programming is a standard approach, it often suffers from high computational complexity and slow convergence.

The authors propose a Path-Following Algorithm framework. The core of this approach involves:

  • Inner Approximations: Replacing complex, non-convex functions with concave quadratic approximations. This transforms the problem into a sequence of simpler convex sub-problems that converge rapidly to a local optimum.
  • Maximin Objective: To ensure fairness and network-wide security, the algorithms maximize the minimum secrecy rate among all users, preventing any single user from being left vulnerable.
  • Robust and Partial CSI Modeling: For partial CSI, statistical guarantees are used. For robust optimization, the authors incorporate outage-probability specifications, assuming channel estimation errors follow an exponential distribution to define a "trust region" for security.

Results and Discussion

The proposed algorithms were evaluated using Monte Carlo simulations of a network with 10 transmitter-user pairs. Key findings include:

  • Convergence and Efficiency: The path-following method demonstrates superior speed, typically converging in under 14 iterations for throughput and 35 for energy efficiency, making it suitable for real-time power control in dynamic networks.
  • The Energy Efficiency Ceiling: Simulations reveal a saturation point where increasing transmit power no longer improves SEE. Eventually, the power consumption in the denominator grows faster than the secrecy gains in the numerator, leading to a decline in efficiency.
  • CSI Insights: Interestingly, the "Partial CSI" model sometimes shows higher "outage throughput" than the "Perfect CSI" model in simulations. This is attributed to the optimistic nature of statistical guarantees compared to the deterministic constraints of the perfect model.
  • Interference Management: The algorithms effectively treat interference as a tool, managing power levels such that interference degrades the eavesdropper's reception more significantly than that of the intended users.

Conclusion

This research provides a robust framework for secure and energy-efficient power allocation in wireless networks. By moving beyond the assumption of perfect CSI and utilizing efficient path-following algorithms, the proposed methods offer a practical path toward securing multi-user interference networks. Future work will focus on extending these techniques to Multi-Input Multi-Output (MIMO) systems and beamforming to further enhance spatial security.

Summary updated

BibTeX
@article{sheng2018power,
  author = {Zhichao Sheng and Hoang Duong Tuan and Ali Arshad Nasir and Trung Quang Duong and H. Vincent Poor},
  title = {{Power Allocation for Energy Efficiency and Secrecy of Wireless Interference Networks}},
  journal = {IEEE Trans. Wireless Communications},
  year = {2018},
  doi = {10.1109/twc.2018.2815626},
}
Figure