Abstract
As wireless networks evolve toward 5G and 6G, energy efficiency has become a critical design constraint. This paper investigates the use of Reconfigurable Intelligent Surfaces (RIS) for downlink multi-user communication. We develop energy-efficient designs that jointly optimize base station transmit power and RIS phase shifts while ensuring individual link budget guarantees. We propose two computationally efficient algorithms: one combining gradient descent with fractional programming, and another utilizing sequential fractional programming. Our results, based on a realistic power consumption model and outdoor propagation environments, demonstrate that RIS-based systems can achieve up to 300% higher energy efficiency compared to traditional multi-antenna amplify-and-forward (AF) relaying.
The Challenge: Sustainable High-Speed Connectivity
The demand for massive data rates in next-generation networks often leads to unsustainable energy consumption. Traditionally, Amplify-and-Forward (AF) relays are used to extend coverage, but their reliance on active power amplifiers makes them energy-intensive.
Reconfigurable Intelligent Surfaces (RIS) offer a passive alternative. An RIS consists of numerous small reflecting elements that can be programmed to alter the phase of incoming signals, effectively "steering" waves toward users without the need for power-hungry amplifiers. However, optimizing these surfaces alongside base station power presents a complex, non-convex mathematical challenge.
System Model and Algorithmic Framework
The study considers a multi-antenna base station serving multiple users via an RIS. To maximize the "bits-per-Joule" performance, the authors employ an Alternating Maximization framework, iteratively solving for transmit power and RIS phase shifts until convergence.
Two primary algorithmic approaches are proposed:
- Gradient Descent (GD) + Fractional Programming: Employs a gradient-based search for RIS phase coefficients and Dinkelbach’s algorithm for optimal power allocation.
- Sequential Fractional Programming (SFP): Approximates non-convex RIS constraints into a sequence of manageable sub-problems, offering slightly superior performance.
Crucially, the paper introduces a realistic power model that accounts for the hardware dissipation of each reflecting element based on its phase-resolution (e.g., 3-bit vs. 6-bit), moving beyond idealized "zero-power" assumptions.
Realistic Simulation and Performance Evaluation
To validate these methods, the authors simulated a realistic outdoor environment using 3GPP propagation models. The setup includes obstacles that block the direct path between the base station and users, necessitating the use of the RIS for reliable communication.
Key Results and Insights
- 300% Efficiency Gain: RIS-based resource allocation provided up to 3x higher energy efficiency compared to traditional AF relays in outdoor scenarios.
- The "Sweet Spot" for RIS Size: Energy efficiency does not scale infinitely with the number of RIS elements ($N$). Because each element adds static hardware power, there is an optimal $N$ beyond which efficiency declines.
- Power Saturation: Unlike spectral efficiency, energy efficiency eventually saturates. Beyond a certain point, increasing transmit power yields diminishing returns in data rate that do not justify the extra energy cost.
- Robustness: The proposed algorithms maintained high feasibility rates (over 99%) even under strict Quality of Service (QoS) requirements.
Conclusion and Future Work
While RIS technology shows immense promise for green networking, several challenges remain. The current study assumes perfect Channel State Information (CSI), which is difficult to obtain for passive surfaces. Future research will focus on channel estimation techniques, the impact of hardware impairments, and performance in highly mobile environments.
Summary updated