Combining Lyapunov Optimization With Evolutionary Transfer Optimization for Long-Term Energy Minimization in IRS-Aided Communications

Pei-Qiu Huang, Yong Wang, Kezhi Wang, Qingfu ZhangIEEE Trans. Cybernetics 2022 · DOI

Abstract

This paper introduces LETO, a framework designed to minimize long-term energy consumption in Intelligent Reflecting Surface (IRS)-aided wireless networks. By integrating Lyapunov optimization for data queue stability with Evolutionary Transfer Optimization (ETO) for accelerated real-time decision-making, LETO addresses the challenges of dynamic user mobility and stochastic data arrivals. Experimental results demonstrate that LETO effectively stabilizes data queues and outperforms traditional optimization methods in energy efficiency and computational speed.

Introduction

Intelligent Reflecting Surfaces (IRS) have emerged as a transformative technology for 6G wireless communications, offering the ability to reconfigure the wireless propagation environment through "smart mirrors." However, deploying IRS in practical, dynamic scenarios presents significant hurdles. Specifically, systems must maintain long-term stability amidst user mobility and unpredictable data bursts while meeting stringent real-time processing constraints.

Traditional "myopic" optimization—focusing solely on the current time slot—often leads to unstable data queues and system failure over time as it ignores the long-term impact of current decisions. Conversely, standard global optimization algorithms are frequently too slow to find high-quality solutions within the millisecond-level windows required for real-time IRS phase-shift updates.

System Model and Methodology

The proposed LETO (Lyapunov Evolutionary Transfer Optimization) framework addresses these issues through a two-stage approach in a simulated IRS-aided uplink environment.

Figure 1: System model of an IRS-aided uplink communication network featuring mobile devices (MDs), an access point (AP), and obstacles.
  1. Lyapunov Optimization: To ensure long-term stability, the complex stochastic problem is decomposed into deterministic sub-problems for each time slot. A "drift-plus-penalty" function is employed to balance immediate energy minimization with the maintenance of short data queues. This framework is particularly powerful because it does not require prior knowledge of future channel states or data arrival patterns.
  2. Evolutionary Transfer Optimization (ETO): Each sub-problem is solved using Differential Evolution (DE). To meet real-time requirements, LETO utilizes ETO to "warm-start" the optimization process. By mapping the similarity of current user locations to past scenarios, the algorithm seeds its initial population with high-quality historical solutions. This knowledge transfer allows the algorithm to converge on optimal solutions much faster than starting from scratch.

Key Insights and Performance

A critical finding of this research is the effectiveness of spatial correlation. In mobile networks, users frequently return to similar spatial configurations. Using physical location as a proxy for problem similarity allows for efficient "knowledge transfer" without the overhead of complex neural-network-based learners, which often struggle with the highly non-linear relationship between location and optimal phase shifts.

Figure 5: Comparison of average queue length over time between myopic optimization and the proposed LETO framework, demonstrating queue stability.

Simulation results in a dynamic environment with stochastic data arrivals highlight LETO's advantages:

  • Queue Stability: As shown in Figure 5, while myopic optimization leads to unbounded queue growth (system failure), LETO successfully stabilizes the average queue length (maintaining it around 0.6–1.0 MB).
  • Scalability: Performance gains over standard DE (LDE) become more pronounced as the number of IRS elements ($N$) increases, proving the transfer method's value for high-dimensional problems.
  • Real-Time Feasibility: LETO achieves decision times between 0.014 and 0.018 seconds, well within the 1-second time slot duration required for practical deployment.

Conclusion and Future Work

LETO provides a robust and efficient solution for energy minimization in IRS-aided networks. By combining the theoretical guarantees of Lyapunov optimization with the speed of evolutionary transfer, it ensures both system stability and real-time feasibility. Future research will explore more complex "multi-hop" environments where signals reflect off multiple surfaces and investigate the impact of localization noise on the transfer mechanism's effectiveness.

Summary updated

Abstract

This article studies an intelligent reflecting surface (IRS)-aided communication system under the time-varying channels and stochastic data arrivals. In this system, we jointly optimize the phase-shift coefficient and the transmit power in sequential time slots to maximize the long-term energy consumption for all mobile devices while ensuring queue stability. Due to the dynamic environment, it is challenging to ensure queue stability. In addition, making real-time decisions in each short time slot also needs to be considered. To this end, we propose a method (called LETO) that combines Lyapunov optimization with evolutionary transfer optimization (ETO) to solve the above optimization problem. LETO first adopts Lyapunov optimization to decouple the long-term stochastic optimization problem into deterministic optimization problems in sequential time slots. As a result, it can ensure queue stability since the deterministic optimization problem in each time slot does not involve future information. After that, LETO develops an evolutionary transfer method to solve the optimization problem in each time slot. Specifically, we first define a metric to identify the optimization problems in past time slots similar to that in the current time slot, and then transfer their optimal solutions to construct a high-quality initial population in the current time slot. Since ETO effectively accelerates the search, we can make real-time decisions in each short time slot. Experimental studies verify the effectiveness of LETO by comparison with other algorithms.
BibTeX
@article{huang2022combining,
  author = {Pei-Qiu Huang and Yong Wang and Kezhi Wang and Qingfu Zhang},
  title = {{Combining Lyapunov Optimization With Evolutionary Transfer Optimization for Long-Term Energy Minimization in IRS-Aided Communications}},
  journal = {IEEE Transactions on Cybernetics},
  year = {2022},
  doi = {10.1109/tcyb.2022.3168839},
}
Figure