Energy-Efficient Irregular RIS-aided UAV-Assisted Optimization: A Deep Reinforcement Learning Approach
Mahmoud M. Salim, Khaled M. Rabie, Ali H. Muqaibel · 2025
Read Energy-Efficient Irregular RIS-aided UAV-Assisted Optimization: A Deep Reinforcement Learning Approach by Mahmoud M. Salim, Khaled M. Rabie, Ali H. Muqaibel (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-15031.txt.
Introduction
Mahmoud Salim and co-authors optimize, in a 2025 paper, a drone-carried smart surface that bounces base station signals to users while harvesting its own energy. Reconfigurable intelligent surfaces can widen coverage and steer beams, but a flying one drains its battery fast. The authors let parts of the surface switch off and draw power from radio waves plus renewable sources. Their strategy splits time between harvesting and signaling to keep service quality within bounds.
The joint problem, spanning transmit power, surface phases, time split, and element choice, is too tangled for classic solvers. A deep reinforcement learning agent, tuned to avoid value overestimates, learns the settings instead. Simulations show harvesting efficiency climbing far above baselines in single and multi-user cases. The authors present the design as a path to longer-lived aerial relays.
Catalog record
| CREATOR | Mahmoud M. Salim, Khaled M. Rabie, Ali H. Muqaibel |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-15031
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Energy-Efficient Irregular RIS-aided UAV-Assisted Optimization: A Deep Reinforcement Learning Approach Mahmoud M. Salim, Khaled M. Rabie, Ali H. Muqaibel Abstract Reconfigurable intelligent surfaces (RISs) enhance unmanned aerial vehicles (UAV)-assisted communication by extending coverage, improving efficiency, and enabling adaptive beamforming. This paper investigates a multiple-input single-output system where a base station (BS) communicates with multiple single-antenna users through a UAV-assisted RIS, dynamically adapting to user mobility to maintain seamless connectivity. To extend UAV-RIS operational time, we propose a hybrid energy-harvesting resource allocation (HERA) strategy that leverages the irregular RIS ON/OFF capability while adapting to BS-RIS and RIS-user channels. The HERA strategy dynamically allocates resources by integrating non-linear radio frequency energy harvesting (EH) based on the time-switching (TS) approach and renewable energy as a complementary source. A non-convex mixed-integer nonlinear programming problem is formulated to maximize EH efficiency while satisfying quality-of-service, power, and energy constraints under channel state information and hardware impairments. The optimization jointly considers BS transmit power, RIS phase shifts, TS factor, and RIS element selection as decision variables. To solve this problem, we introduce the energy-efficient deep deterministic policy gradient (EE-DDPG) algorithm. This deep reinforcement learning (DRL)-based approach integrates action clipping and softmax-weighted Q-value estimation to mitigate estimation errors. Simulation results demonstrate that the proposed HERA method significantly improves EH efficiency, reaching up to 81.5% and 73.2% in single-user and multi-user scenarios, respectively, contributing to extended UAV operational time. Additionally, the proposed EE-DDPG model outperforms existing DRL algorithms while maintaining practical computational complexity. I Introduction The sixth generation (6G) wireless communication system has been driven by the increasing demand for seamless connectivity, high data rates, and low-latency services [ 1 , 2 ] . With the rapid proliferation of connected devices, including user equipment, Internet of Things (IoT) devices, and autonomous systems, future networks should address critical challenges such as spectrum limitations, energy efficiency (EE), and link reliability. Ensuring robust and adaptive network coverage is essential for various applications, including real-time data transmission, remote sensing, and intelligent automation. Unmanned aerial vehicles (UAVs) have emerged as a transformative solution to address communication challenges, offering high mobility, flexible deployment, and rapid response capabilities [ 3 ] . They play a critical role in public safety scenarios, where conventional infrastructure is often unreliable or unavailable. By leveraging line-of-sight (LoS) communication and rapid deployment, UAVs can function as wireless relays, mobile base stations (BSs), or aerial network extenders, significantly improving connectivity in complex environments. However, UAV-assisted communications are highly susceptible to blockages from buildings and terrain, signal attenuation, and energy constraints, which limit their operational time and long-term effectiveness. To enhance their communication reliability, UAVs have been integrated with reconfigurable intelligent surfaces (RISs), creating programmable radio environments by dynamically controlling signal reflections [ 4 ] . Unlike conventional relays, passive RIS operates without amplification, improving spectral efficiency (SE) and EE without requiring additional radio frequency (RF) chains [ 5 , 6 ] . UAV-mounted RISs provide full-space reflection and adaptive coverage, effectively mitigating signal blockages, extending communication range, and optimizing energy consumption. Energy harvesting (EH) can play a vital role in extending the UAV-mounted RIS trip time by incorporating both RF and renewable energy (RE) sources [ 4 , 7 , 8 ] . While RF EH provides a continuous but limited energy supply, RE sources generate power intermittently and can significantly contribute when available. By leveraging both energy sources, UAV-mounted RIS can enhance operational sustainability, reduce reliance on onboard batteries, and extend mission duration in energy-constrained environments. Optimizing UAV-mounted RIS systems is challenging due to the highly coupled control variables, including UAV trajectory, RIS phase-shift design, and power allocation [ 9 ] . The dynamic nature of UAV movement and the need for real-time adaptation to channel variations add further complexity. Several optimization approaches have been explored. Model-based methods such as alternating optimization [ 10 ] , successive convex approximation (SCA) [ 11 ] , and semidefinite relaxation [ 12 ] offer structured solutions but often involve high computational complexity. Heuristic techniques provide lower complexity alternatives but may compromise solution optimality [ 13 ] . More recently, machine learning-based approaches, particularly deep reinforcement learning (DRL), have gained attention for handling complex, dynamic environments and large solution spaces efficiently [ 14 ] . By leveraging DRL algorithms such as deep deterministic policy gradient (DDPG) and its variants, UAV-mounted RIS systems can dynamically adjust trajectory, beamforming, and RIS configurations. This enables improved SE, reduced power consumption, and enhanced overall network performance. I-A Related Work RIS improves wireless network performance by reconfiguring signal propagation to enhance energy and SE. In [ 15 ] , the authors analyzed the joint design of transmit beamforming and RIS phase shifts in a multiple-input multiple-output (MIMO) multi-user (MU) system using DRL to optimize the sum-rate of the users. In [ 6 ] , the paper studied the SE-EE trade-off in a RIS-assisted full-duplex MU MIMO system using a DRL-based approach. The RIS phase shifts and power allocation were jointly optimized while considering partial channel state information (CSI), co-channel interference, and residual self-interference to enhance system performance. The work in [ 16 ] examined a RIS-assisted multiple-input single-output (MISO) symbiotic radio system with finite blocklength transmissions. An EE maximization problem was formulated by jointly optimizing the BS beamforming vector and RIS phase shifts, and a DRL-based framework was proposed to solve the non-convex problem. Additionally, a distributed RIS architecture was explored to improve EE by coordinating the joint optimization of transmit power and RIS scheduling using SCA [ 17 ] . The integration of RIS with EH enhances EE in wireless networks by optimizing EH and management. Studies focus on different aspects including power control, phase shift optimization, and system efficiency. The work in [ 18 ] optimized an active RIS-assisted downlink system with time-switching (TS) EH. It jointly configured RIS phase shifts, time allocation, and power control using a two-step convex optimization approach to enhance EE. In [ 19 ] , the authors investigated a system of RIS-assisted massive MIMO enabling wireless power transfer to energy-limited IoT devices. The BS transmit power and RIS reflection coefficients were optimized utilizing SCA, considering CSI errors to enhance harvested energy. In [ 20 ] , the authors studied a passive RIS-assisted device-to-device cognitive radio network, where a batteryless secondary transmitter harvested energy from RF signals. They maximized EE by jointly optimizing transmit power and RIS beamforming via an alternating optimization algorithm. The work in [ 7 ] analyzed a wireless-powered IoT network assisted by passive and active RIS, considering a nonlinear EH model. They derived analytical expressions for outage probability, sum throughput, and EE. In [ 21 ] , the authors optimized power splitting and transmit power in RIS-assisted simultaneous wireless information and power transfer (SWIPT) systems. This came with a phase-dependent amplitude model for RIS reflectivity employing fractional programming-based solutions to maximize data rate and EH. The RIS-assisted SWIPT networks were investigated in [ 22 ] , formulating an EE maximization problem considering quality-of-service (QoS) constraints. A proximal policy optimization (PPO)-based DRL approach was proposed to jointly optimize beamforming, power splitting, and discrete phase shifts. In [ 8 ] , a dynamic adaptable EH model was integrated into the RIS, allowing it to harvest wireless energy from various sources and improve