A Comprehensive Review of Energy-Efficient Techniques for UAV-Assisted Industrial Wireless Networks

Abstract

The rapid expansion of the Industrial Internet-of-Things (IIoT) has spurred significant research interest due to the growth of security-aware, vehicular, and time-sensitive applications. Unmanned aerial vehicles (UAVs) are widely deployed within wireless communication systems to establish rapid and reliable links between users and devices, attributed to their high flexibility and maneuverability. Leveraging UAVs provides a promising solution to enhance communication system performance and effectiveness while overcoming the unprecedented challenges of stringent spectrum limitations and demanding data traffic. However, due to the dramatic increase in the number of vehicles and devices in the industrial wireless networks and limitations on UAVs’ battery storage and computing resources, the adoption of energy-efficient techniques is essential to ensure sustainable system implementation and to prolong the lifetime of the network. This paper provides a comprehensive review of various disruptive methodologies for addressing energy-efficient issues in UAV-assisted industrial wireless networks. We begin by introducing the background of recent research areas from different aspects, including security-enhanced industrial networks, industrial vehicular networks, machine learning for industrial communications, and time-sensitive networks. Our review identifies key challenges from an energy efficiency perspective and evaluates relevant techniques, including resource allocation, UAV trajectory design and wireless power transfer (WPT), across various applications and scenarios. This paper thoroughly discusses the features, strengths, weaknesses, and potential of existing works. Finally, we highlight open research issues and gaps and present promising potential directions for future investigation.

1. Introduction

1.1. Background

In recent years, investigations on the Industrial Internet of Things (IIoT) have witnessed increasing popularity driven by the rapid expansion of its market worldwide. Statista projects that the IIoT market segment is expected to rise from 238.40 billion US$ to 454.90 billion US$ between 2024 and 2029, particularly in China [1]. As an integrated part of the broader Internet-of-Things (IoT), the IIoT typically aims at connecting a large number of wireless devices together at a reasonable cost. Prominent IIoT applications include smart cities [2], intelligent transport systems (ITS) [3] consisting of a large number of vehicles and devices to collect and process massive data across networks [4]. Hence, quality-of-service (QoS) constraints such as security and time-sensitive requirements are considered to ensure system reliability [5]. However, the dramatic surge in data requirements leads to a rapid rise in the number of devices deployed in industrial wireless networks. Additionally, vehicles are no longer just transportation tools but are increasingly being exploited as operational platforms. Hence, employing unmanned aerial vehicles (UAVs) as aerial base stations offers a viable solution to efficiently and reliably implementing industrial wireless networks due to their high mobility, flexibility, and maneuverability [6]. Additionally, UAVs are leveraged to serve users in remote areas for network expansion, significantly reducing the infrastructure construction costs such as those associated with terrestrial base stations (TBSs). Moreover, UAVs facilitate the establishment of wireless power transfer among devices deployed in IIoT systems, promoting sustainable network operation while overcoming energy limitations. Furthermore, utilizing UAVs can establish line-of-sight (LoS) communication links between devices and vehicles, facilitating real-time data transfer and supporting the sustainability of the IIoT system [7].

In practice, two types of UAVs are commonly exploited: fixed-wing UAVs, which can achieve high-speed motion and operate at high altitudes, are commonly exploited in surveying, mapping, and monitoring tasks for smart agriculture [8] and disaster managements [9]. On the other hand, rotary-wing UAVs, known for their flexibility and ability to hover at certain positions, are adopted in environmental monitoring systems for surveillance and air quality testing [10], and smart cities [11] to facilitate traffic monitoring and ensure public safety. Thus, with the assistance of UAVs, common challenges faced in industrial wireless networks, such as stringent spectrum limitations and heavy data traffic, can be significantly alleviated [12].

Moreover, due to size and weight limitations of the devices and vehicles [13], UAV-assisted IIoT networks encounter critical constraints such as battery storage [14], operation time [7] and working offloads [15]. Due to stringent constraints on energy consumption, most UAVs cannot continue operating for more than 1 h, although the maximum recorded duration reaches approximately 81 h [16]. As such, overcoming these issues is essential for reliable UAV-assist system operation. To this end, developing and adopting energy-efficient techniques is crucial for IIoT applications to ensure their system sustainability and prolong the network and devices’ lifetime.

To address these challenges, we comprehensively review the techniques for addressing energy-efficient issues in UAV-assisted industrial wireless networks. We begin by outlining the scope of existing energy-efficient technique reviews related to UAV-assisted networks and industrial wireless networks, highlighting our motivations and contributions. Next, we explore energy-efficient techniques in industrial wireless networks, followed by explaining the capabilities of UAVs in IIoT. Subsequently, techniques for ensuring energy efficiency in UAV-assisted networks, such as UAV placement, resource allocation, and wireless power transfer (WPT) are discussed and summarized in detail. We also provide several case studies of UAV-assisted industrial wireless networks that exploit energy-efficient techniques, identify open research challenges and present potential solutions to enhance the performance of UAV-assisted IIoT networks. Finally, we summarize and conclude the paper.

1.2. Related Reviews for Energy-Efficient Techniques on UAV and Industrial Wireless Networks

Over the years, researchers have devoted themselves to investigating and reviewing energy-efficient techniques for enhancing UAV-assisted systems. For instance, Ref. [17] provided a comprehensive survey on energy-efficient UAV-assisted 6G networks. In particular, five key energy-efficient techniques, including UAV placement, energy harvesting, and power allocation, are stated in the paper with detailed explanations for UAV energy consumption models. However, the survey primarily focused on 6G applications and overlooked the potential of the cooperation of UAVs and IIoT applications. Another related survey [18] aimed to research energy optimization techniques for UAV-assisted cellular networks, summarizing existing optimization methodologies and dividing them into two tracks: conventional or machine learning (ML)-based solutions. Likewise, IIoT was not highlighted in this paper. In contrast, instead of presenting algorithmic solutions, we focus on surveying energy-efficient techniques in UAV-assisted industrial wireless networks.

Additionally, Ref. [19] investigated the energy consumption of UAVs associated with their routing. Specifically, the authors reviewed various factors that influence UAVs’ energy consumption, including the weather and parameters of the UAV, such as flying speed and payload. Although Ref. [19] also provides detailed energy consumption models for the UAV, compared with our work, it lacks energy-efficient techniques and ignores the IIoT. On the other hand, from a public safety perspective, Ref. [20] reviewed UAV survey papers from a public safety point of view while considering energy-efficiency issues. Specifically, a multi-layer UAV network is introduced in [20], was focusing on UAV deployment techniques, yet again, it omitted any discussing IIoT. Furthermore, a number of energy-efficiency techniques, protocols and algorithms were reviewed in [16], but IIoT was not discussed together, while other techniques, such as scheduling, are not included in [16], which were important in IIoT applications.

Another line of literature focuses on the industrial wireless network perspective; in [21], the authors highlighted the concept of “smart energy” in IIoT, which includes use cases and renewable energy sources with corresponding networks. Also, a survey prospectively on vehicular intelligence was proposed in [22], where energy-efficiency techniques such as light-emitting diode (LED) lighting installations, mmWaves, and satellite communications are discussed in IIoT applications. Additionally, a survey on federated learning approaches for energy-efficient, green and sustainable IIoT was presented in [23], which focuses on reviewing AI-based algorithms. However, all three IIoT energy-efficiency surveys did not prioritize UAV assistance and mention it only briefly in one or two subsections about UAV/air networks, and failed to provide detailed reviews on energy-efficient techniques. Specifically for UAV-assisted Industrial wireless networks. A list of existing reviews is provided in

Table 1. List of Related Reviews.

Title	Year	Main Achievements	Limitations
[23]	2024	Focused on reviewing AI-based algorithms for energy efficient perspective in industrial wireless networks.	Lack of focus on UAV-assisted networks nor detailed reviews on energy-efficient techniques in UAV-assisted Industrial wireless networks.
[16]	2023	Provided a number of energy-efficiency-related techniques, protocols and algorithms.	Techniques such as scheduling are not included and there is no focus on industrial wireless networks.
[18]	2023	Summarized energy optimization techniques for UAV-assisted cellular networks and existing optimization methodologies.	UAV-assist industrial wireless networks have not been highlighted in this paper.
[21]	2022	Highlighted the concept of “smart energy” in IIoT applications.	Lack of focus on UAV-assisted networks nor detailed reviews on energy-efficient techniques in UAV-assisted Industrial wireless networks.
[22]	2022	Introduced energy-efficiency techniques in IIoT networks such as LED lighting installations, mmWaves and satellite communications.
[17]	2022	Detailed explanations for UAV energy consumption models with five energy-efficient techniques in UAV-assisted 6G networks.	Overlooked the potential of the cooperation of UAVs and IIoT applications.
[19]	2019	Investigated the energy consumption of UAVs associated with their routing.	Lack of energy-efficient techniques and ignores the Industrial wireless network.
[20]	2019	Reviewed public safety UAV paper and addressing energy-efficiency issues.	Insufficient energy-efficient techniques provided and no focus on IIoT-related work.

.

1.3. Motivation and Contribution

According to our literature review, most existing surveys have overlooked the energy-efficient techniques in UAV-assisted industrial wireless networks. Also, there is incompleteness in the coverage of energy-efficient techniques highlighted in the existing literature. To address this gap, we aim to comprehensively review energy-efficient techniques for industrial wireless networks, UAV-assisted networks, and UAV-assisted industrial wireless networks with capabilities of UAVs in industrial wireless networks. Our objectives are two-fold:

(a) We review and analyze existing energy-efficient techniques for industrial wireless networks, UAV-assisted networks, and UAV-assisted industrial wireless networks. We aim to highlight their features, strengths, and weaknesses. Additionally, we also elucidate the capabilities of UAVs in industrial wireless networks.

(b) We discuss and identify features, strengths, weaknesses, and potentials of energy-efficient techniques in the existing literature. We also pinpoint various open fundamental research problems in this rapidly evolving research area to enhance energy-efficient techniques and suggest future directions.

The four-fold contributions of this paper can be summarized as follows:

(1) We provide detailed reviews of energy-efficient techniques in industrial wireless networks and UAV-assisted networks and analyse their features, strengths, weaknesses, and potentials.

(2) We identify UAV’s capabilities, including their roles as aerial base stations (ABSs), network expansion, and cost reduction. We explain the reason for leveraging UAVs in traditional industrial wireless networks and provide energy consumption models for two types of commonly employed UAVs.

(3) We highlight existing energy-efficient UAV-assisted industrial wireless networks and discuss the energy-efficient techniques in this literature.

(4) We articulate open research challenges in UAV-assisted industrial wireless networks, such as developing accurate models and utilizing renewable energy sources. Additionally, we will outline future research directions.

1.4. Paper Organization

The rest of the paper is organized as follows: Section 2 introduces existing energy-efficient techniques in industrial wireless networks, highlighting important approaches such as leveraging vehicular or backscatter communication. Next, in Section 3, we detail the capabilities of UAVs to facilitate IIoT applications and explain their role in enhacing industrial wireless networks. After that, we review and analyze the energy-efficient techniques specifically for UAV-assisted networks employed in existing literature in Section 4. With the understanding of UAV’s capability and explanations of energy-efficient techniques in UAV-assisted networks and industrial wireless networks, we discuss and highlight several existing energy-efficient techniques and present open research problems in Section 5. Finally, we summarize and conclude our work in Section 6.

The structure and main concepts of the paper are shown in Figure 1 for clarity. The three main parts of energy-efficient UAV-assisted industrial wireless networks are “Energy Efficiency”, “Industrial Wireless Networks”, and “UAV Assistance”. Consequently, Section 2 identifies the combination of energy efficiency and industrial wireless networks. Section 3 introduces performance enhancements of industrial wireless networks with UAVs’ assistance. Next, the combination of energy efficiency and UAV communication networks is discussed in Section 4, followed by the analysis of energy-efficient UAV-assisted industrial wireless network provided in Section 5.

2. Energy Efficient Techniques in Industrial Wireless Network

In this section, we provide a comprehensive analysis of the energy-efficient techniques adopted in industrial wireless networks. As mentioned before, with the 5G commercial adopted worldwide, there has been a significant acceleration in the implementation of Industry 4.0 within the Industrial IoT landscape [24]. Hence, the inherent ultra-reliable low latency communications (URLLC) and massive machine-type communications (mMTC) facilitate the IIoT in the real world. However, as IIoT require massive devices connected to the system, energy consumption becomes a critical challenge. Therefore, this section will review the energy-efficient techniques in industrial wireless networks; we start by introducing vehicular networks, then backscattering communication, and WiFi sensing. Finally, we will review relevant surveys and research papers related to this topic.

2.1. Vehicular Communication

Closer and deeper integration between logistics supply chain elements, such as supply, distribution, transport, and warehousing, is enabled by 5G technology, facilitating a better-coordinated strategy with cross-industry integration [25]. In particular, IoT sensors and smart devices can be connected to 5G networks to form a unified platform that facilitates data exchange between different organizations in the logistics process [26]. This will be useful for dynamic information acquisition and tracking goods information [27] for seamless supply chain management, from production to customer delivery. This logistics requirement highlights the importance of vehicular communication.

During the industry $4.0$ era, the seamless connectivity afforded by 5G has been pivotal in advancing vehicular-to-everything (V2X) communications, thereby enabling the feasibility of autonomous driving. This advancement significantly enhances road capacity, improves traffic safety, and contributes to a reduction in energy consumption [28]. Indeed, researchers in [29] have pointed out that 5G-enabled automated guided vehicles (AGVs) are increasingly being deployed on smart shop floors and in smart storage systems. These AGVs not only cooperate with infrastructures and other operational equipment to facilitate the loading and unloading of goods but also work collaboratively with other AGVs. Furthermore, they autonomously locate power sources for recharging. Throughout this process, AGVs are capable of perceiving environmental maps, positions of other AGVs, and their own battery levels, and can make quick calculations and decisions.

2.2. Platooning System

In response to the dramatic increase in vehicle numbers and advancements in vehicular communication, researchers have proposed vehicular platooning systems as a method to enhance road capacity and energy efficiency. These systems accomplish this by reducing inter-vehicle distances and minimizing abrupt accelerations and decelerations. Within the IIoT, platooning utilizes vehicular communication technologies to significantly bolster supply chain operations. These platooning systems not only have the potential to improve road capacity and safety by up to $20\%$ [30], but they also contribute significantly to reductions in energy consumption and carbon dioxide (CO₂) emissions [31]. Research on vehicle platooning, which originated in the 1970s [32], has experienced extensive development globally since the 1990s. The practical application of these systems in real-world scenarios has been largely enabled by the advent of advanced 5G technologies [33].

In practice, vehicle communication enables the process of making vehicle platoons autonomous, resulting in a significant reduction in the distance between vehicles. This reduces aerodynamic drag and enhances the efficiency of road traffic. The advancement of platoon control technology has enabled the implementation of fundamental driving functions and facilitated the comprehensive scheduling and planning of vehicle platoons [34]. This development is crucial for the extensive adoption of industrialized platoon systems, which have the potential to offer substantial economic advantages, especially in supply chain scenarios [35].

2.3. Backscatter Communication

The above-mentioned platooning system joined with vehicular communication, has significantly improved the IIoT performance, but with the massive connectivity demanded by the IIoT, low-energy communication techniques are essential within wireless industrial networks. Hence, utilizing backscatter techniques offers a promising solution for implementing sustainable networks [36] with a low financial cost while ensuring energy efficiency [37]. In this context, we introduce the concept of a green “paradise-backscatter” network [38]. Backscatter techniques, traditionally employed in applications such as radio-frequency identification (RFID) for item tracking and medical telemetry [39], are evolving. In particular, ambient backscatter has emerged as a promising technology. It represents a significant advancement over conventional backscattering methods by enabling low-energy communication systems that overcome traditional communication limitations [40]. In the backscattering framework, the backscatter tag captures the surrounding continuous wave (CW) and modulates its own message in the reflection to the receiver, requiring little or no power supply [41]. This unique capability is particularly valuable in a massively connected environment, such as the IIoT, highlighting its potential for future low-energy communication systems [42].

Due to their high energy efficiency, backscattering techniques have been extensively investigated by several pioneers in industrial networks [43]. These techniques have also been integrated with other technologies, such as vehicular communication. As illustrated in Figure 2, backscatter techniques can be adopted within the V2X framework to help lower energy consumption. However, this integration introduces a non-trivial trade-off, as it raises concerns regarding data transmission security, such as eavesdropping attacks. For instance, the authors in [44] established a backscatter signal model for the chipless tag and provided insights into its performance when adopted in real environments. Furthermore, Ref. [45] highlighted that the next generation of the industry will focus on creating an ecosystem for green transportation with minimal carbon dioxide emissions. In this context, a green wireless communication technique, namely backscatter networks, was poised to be a key enabler for the next generation of industry. In fact, backscatter techniques could also help healthcare networks in the IIoT, with applications ranging from on-body sensors to in-body implants and small devices that can be embedded [46].

2.4. Reconfigurable Intelligent Surfaces

Similar to the energy-efficient feature of backscatter technology, reconfigurable intelligent surfaces (RIS) have gained considerable interest due to their exceptional capability to improve communication coverage by manipulating the phase shift of individual reflecting elements instead of depending on power amplifiers [47]. In particular, RIS offers additional advantages, such as ease of implementation, improved spectral efficiency, eco-friendliness, and compatibility [48]. Because RIS devices are generally passive, consisting of electromagnetic materials. they can be seamlessly integrated into various structures, such as building facades, highway poles, vehicle windows, and even clothing [49]. More importantly, RIS, with its ability to modify the wireless propagation environment, efficiently mitigates power attenuation over extended distances and effectively resolves signal obstruction problems.

Due to the stringent energy efficiency requirements of the new generation of industry, Jamil et al. focused on resource allocation to maximize the energy efficiency of RIS-assisted wireless networks in industrial scenarios [50]. Additionally, Dhok et al. proposed a RIS-assisted wireless communication framework incorporating non-linear energy harvesting for industrial automation [51]. This study examined the proposed system’s reliability performance and highlighted the trade-off between bits per channel use and channel uses. Furthermore, an innovative infrastructure based on shape-adaptive RIS, suitable for smart IIoT, was proposed in [52]. The goal of this infrastructure is to control RIS-reflected waves in the shape domain. This can be achieved by making RIS unit sizes more flexible and physical shapes that can be changed on a large scale. More recently, Aboagye et al. explored the potential of RIS in high-bandwidth visible light communication (VLC) systems, enhancing the opportunities for adopting VLC in real-world applications [53].

2.5. WiFi Sensing

Recently, the use of wireless sensing technology is revolutionizing Wi-Fi devices, converting them into omnipresent sensors, and greatly enhancing Wi-Fi systems’ functionalities, capacities, and applications. This change is fundamentally altering the sensing process, particularly in the context of human-centric sensing. Wi-Fi sensing utilizes ambient Wi-Fi signals to analyze and interpret the movements of individuals and objects, enabling various applications such as motion sensing, sleep monitoring, and fall detection [54]. It has also been recognized that Wi-Fi sensing is suitably adapted for monitoring and counting tasks within the IIoT. Although traditional surveillance technologies such as cameras or ultrasonic sensors are efficient, they frequently give rise to privacy problems and lack the widespread availability for quick implementation. Wi-Fi sensing is a favourable alternative in this situation since it provides a solution without the expenses and privacy issues that come with traditional methods [55].

2.6. Simultaneous Wireless Information and Power Transfer

The ever-growing number of wirelessly connected devices has significantly increased energy consumption, making alternative wireless information and power transfer techniques critical not only for theoretical research but also for reducing operational costs and supporting the long-term growth of wireless communications. In this context, radio-frequency (RF) energy harvesting presents a transformative paradigm for effective wireless communication systems, enabling wireless nodes to recharge their batteries adopting RF signals instead of relying on fixed power grids or traditional energy sources [56]. By letting power and information be sent simultaneously, simultaneous wireless information and power transfer (SWIPT) technologies offer significant advantages in terms of power use, spectrum efficiency, interference management, and transmission delay [57]. Furthermore, Tang et al. proposed a joint power allocation and time-switching control algorithm to maximize energy efficiency with SWIPT [58].

Furthermore, the authors in [59] investigated the energy-efficient optimization problem for device-to-device (D2D) communications within UAV-assisted IIoT networks featuring SWIPT. Inspired by the industrial revolution integrating the physical and informatics realms, the first survey paper providing an overview of recent design trends in SWIPT systems within the transportation and industrial sectors was published in [60] in 2021. Recent research also indicates that SWIPT can enhance wireless federated learning by facilitating over-the-air wireless power transfer concurrently with the transmission of global and local models [60]. This capability allows edge devices to conduct local machine learning training while adhering to data privacy and meeting energy constraints. Simultaneously, SWIPT has demonstrated potential in assisting the optimization of energy-efficient cooperative IIoT networks [61].

3. UAV Assistance in Industrial Wireless Networks

This section explores the potential of utilising UAVs in industrial wireless networks. Building on the background previously discussed, UAVs posses high mobility and controllability, which could significantly improve the performance of existing IIoT applications. Here, we offer a detailed review and explain how UAVs facilitate industrial wireless networks and contribute to revamping their weaknesses, including serving as ABSs, expanding network coverage, cost reduction, and achieving wireless power transfer. Since wireless power transfer will be reviewed in detail in other sections, we compare two types of commonly leveraged UAVs in industrial wireless networks and analyse corresponding energy consumption models.

3.1. Aerial Server

Due to UAVs’ high mobility and maneuverability, employing them as ABSs serves as a promising method to improve the performance of wireless networks while ensuring reliability and effectiveness. Traditional industrial wireless networks consist of a large number of devices that have urgent limitations on resources such as spectrum, power, and channel/bandwidth. Adopting UAVs as ABSs and carrying servers to compute real-time processing on resource allocation can significantly overcome the common challenges in industrial wireless networks, as mentioned above. For instance, ML-based resource allocation solutions such as deep reinforcement learning (DRL) can be implemented on the UAVs under multi-UAV [62] or mmWave scenario [63]. Also, it can assist ultra-dense networks (UDN), which serve a large number of users. Additionally, to address the computation burden in IIoT networks, especially those with stringent resource constraints, adopting mobile-access edge computing (MEC) by utilizing UAVs is a promising method [64]. Also, in extreme cases such as TBSs or IIoT infrastructures are paralysed due to disasters and cannot work quickly, UAVs can serve as ABSs and release the burden of industrial wireless networks [65].

Another benefit of deploying ABS in industrial wireless networks is improving the quality of the wireless channels by establishing LoS domain communication links between the UAV and the devices or users [66]. The dramatic increase in the number of users and devices of industrial wireless networks presents significant challenges to transmitting and processing data ahe network. Establishing LoS connectivity is extremely important for fast and reliable wireless communication [67]. Hence, employing UAVs as ABSs and utilizing them as servers while ensuring LoS communication links across the system can significantly address the existing issues of industrial wireless networks.

Moreover, IIoT applications face stringent constraints related to time sensitivity [68] since most require collecting and processing data on the server and making decisions as fast as possible to capture the feature of real-time network reactions to various events. To overcome these challenges, employing UAVs as ABSs to facilitate industrial wireless networks is practical since ABSs equipped with powerful computing units can process collected data and return the results to devices in real-time, which significantly reduces data latency even compared to transmitting data between the device and a cloud server [69]. For example, deploying ABSs and utilizing MEC is a potential solution to address computation and latency-aware missions [70] in networks with heavy data traffic. As a result, exploiting UAVs as ABSs assigned with different roles can achieve URLLCs and support mMTC [71], which can notably enhance latency, energy efficiency, and connection probability.

3.2. Coverage Enhancement

Conventional industrial wireless networks consist of TBSs and devices. Hence, techniques for effectively connecting while ensuring the coverage of TBSs, devices, and users are essential and challenging in current IIoT applications. Also, IoT devices become ineffective when deployed in places without TBSs’ coverage [72]. To address this issue, leveraging and placing UAVs flexibly becomes a promising, cost-effective approach to expanding the network coverage in a high-dimensional space and wider area [73]. Owing to their strong mobility and ability to improve wireless channels, UAVs can provide sufficient service coverage with no gross distortion such as severe path loss [74], pertaining to applications adopting high-level techniques such as non-orthogonal multiple access (NOMA) and MEC [75]. Moreover, exploiting UAVs enabled with edge servers and smartly optimizing their trajectory can implement edge computing in the existing industrial wireless networks and increase the size of the service area [62].

Furthermore, utilizing UAVs is a practicable method to support rural applications, such as smart farms, military operations, and IIoT devices located outside the coverage of cellular systems, which cannot be served by edge computing and MEC [76]. Additionally, UAVs can further enhance the network coverage, attempt to liaise with users within the disaster area, and facilitate a prompt response to first aid operations [77].

3.3. Cost Reduction

As a result of the high financial and construction costs associated with deploying TBSs and static servers [78] in industrial wireless networks, and the dramatic development of IIoT, a large serving or coverage area is required with new TBSs and static servers constructed where employing UAV is a promising and practical method to reduce system building expenses. UAVs are proven to be low-cost solutions capable of realizing intelligent management [79], and leveraging small UAVs can even further reduce the cost while satisfying the requirements and providing reliable connection across the device deployed in industrial wireless networks [80].

Additionally, by exploiting UAVs equipped with edge servers and strategically optimizing their trajectory, edge computing can be achieved even without cooperating with other wireless infrastructures [62]. Also, it is worth noting that exploiting fog UAV wireless networks offers higher flexibility than utilizing TBSs due to UAVs’ swift deployment and low construction costs [81]. In other words, owing to their great benefits, UAV-assisted industrial networks are becoming the future trend in IIoT applications.

3.4. UAV and Energy Consumption Models

There are two common wireless power transfer methods on the UAV side: let UAVs charge the devices by exploiting their battery storage [82] or let UAVs charge but also allow them to harvest energy from other devices [83] (e.g., TBSs and other UAVs). Since most of the devices deployed in current industrial wireless networks have limited battery storage, utilizing UAVs to wirelessly power the other devices has drawn interest and become popular in recent IIoT research [82,83,84,85,86]. However, UAVs have stringent energy and operation time constraints due to their limited size and weight. Hence, applying energy-efficient techniques is crucial in UAV-assisted industrial wireless networks, which will be explained in detail in later sections. Here, we first explain and compare different types of UAVs and corresponding energy consumption models. In the beginning, we highlight that communication-related energy is often ignored in the existing energy-efficient UAV papers [87] since it is relatively low compared to operational energy consumption. Hence, in this section, we focus on explaining the operational energy consumption models.

3.4.1. Fixed-Wing UAV

Fixed-wing UAVs have a similar structure to aeroplanes, as they rely on forward motion and aerodynamic lift from their wings to fly. This kind of UAV has the ability to move at high speed and altitude between 1 to 10 km [88], which can be exploited in applications requiring long-distance travel and extended flight times, such as mapping [89], surveying [90], agriculture [91], and surveillance [92]. Due to their high speed, considering the Doppler effect is essential in fixed-wing UAV-assisted systems, some existing papers, e.g., [93], assume that the receivers of the UAV and IoT nodes can perfectly compensate for it. Also, launching or landing fixed-wing UAVs requires a runway or a catapult system, and a vertical takeoff and landing (VTOL) strategy has to be adopted. Additionally, fixed-wing UAVs’ flying direction is limited since they cannot move backwards [94].

For a fixed-wing UAV with mass and the gravitational constant given by $m_{\mathrm{fw}}$ and g, that flies at $v_{\mathrm{fw}}\left(t\right)$ instantaneous speed with acceleration of $a_{\mathrm{fw}}\left(t\right)$ at timestamp t, the total propulsion energy $E_{\mathrm{fw}}$ during T time duration is denoted as [95]:

$$\begin{array}{cc}\hfill E_{\mathrm{fw}}\left(v_{\mathrm{fw}}\left(t\right),a_{\mathrm{fw}}\left(t\right)\right)=& {\int }_0^T\left[c_1{v}_{\mathrm{fw}}{\left(t\right)}^3+{\displaystyle \frac{c_2}{v_{\mathrm{fw}}\left(t\right)}}\left(1+{\displaystyle \frac{a_{\mathrm{fw}}{\left(t\right)}^2-\frac{{\left(a_{\mathrm{fw}}^T\left(t\right)v_{\mathrm{fw}}\left(t\right)\right)}^2}{v_{\mathrm{fw}}{\left(\right)}^2}}{g^2}}\right)\right]dt\hfill \\ \hfill & +{\displaystyle \frac{1}{2}}{m}_{\mathrm{fw}}\left(v_{\mathrm{fw}}{\left(T\right)}^2-v_{\mathrm{fw}}{\left(0\right)}^2\right)\hfill \end{array}$$

(1)

where $c_1$ and $c_2$ are the parameters related to UAV’s weight, wing area and air density.

3.4.2. Rotary-Wing UAV

Compared with fixed-wing UAVs, rotary-wing UAVs are similar to helicopters, which generate thrust force from rotors. Although rotary-wing UAVs cannot achieve high-speed movement as their fixed-wing counterparts, they are highly maneuverable and can move in any direction, hover at a certain position and achieve VTOL. The applications of rotary-wing UAVs include search and rescue systems [96], infrastructure inspection and environmental monitoring networks [97]. The high flexibility of rotary-wing UAVs sacrifices their flight times and ranges due to their less efficient aerodynamics and higher power consumption [98].

For a rotary-wing UAV that flies at $v_{\mathrm{rw}}\left(t\right)$ instantaneous speed with acceleration of $a_{\mathrm{aw}}\left(t\right)$ at timestamp t, we first define the thrust force $T_{\mathrm{rw}}(v_{\mathrm{rw}}\left(t\right),a_{\mathrm{rw}}\left(t\right))$ as [99]:

$$T_{\mathrm{rw}}(v_{\mathrm{rw}}\left(t\right),a_{\mathrm{rw}}\left(t\right))={\displaystyle \frac{1}{n_r}}\left(m_{\mathrm{rw}}{a}_{\mathrm{rw}}\left(t\right)+{\displaystyle \frac{1}{2}}\rho v_{\mathrm{rw}}\left(t\right)S_{\mathrm{FP}}\right)v_{\mathrm{rw},\mathrm{d}}\left(t\right)-m_{\mathrm{rw}}g,$$

(2)

where $n_{\mathrm{r}}$ and $\rho$ represent the rotor number and air density, respectively, and the mass of the UAV and the gravitational constant is given by m and g. $S_{\mathrm{FP}}$ is the fuselage equivalent flat plate area, and the velocity direction vector is denoted as $v_{\mathrm{rw}},\mathrm{d}\left(t\right)$. The power that generates the thrust is denoted as [99]:

$$\begin{array}{cc}\hfill P_{\mathrm{rw}}(T_{\mathrm{rw}},v_{\mathrm{rw}}\left(t\right))=n_{\mathrm{r}}& \left[{\displaystyle \frac{\delta }{8}}\left({\displaystyle \frac{T_{\mathrm{rw}}}{c_{\mathrm{T}}\rho A}}+3v_{\mathrm{rw}}{\left(t\right)}^2\right)\sqrt{{\displaystyle \frac{\rho c_{\mathrm{s}}^{2}A{T}_{\mathrm{rw}}}{c_{\mathrm{T}}}}}\right.\hfill \\ \hfill & +{\left(\sqrt{{\displaystyle \frac{T_{\mathrm{rw}}^2}{4{\rho }^2{A}^2}}+{\displaystyle \frac{v_{\mathrm{rw}}{\left(t\right)}^4}{4}}}-{\displaystyle \frac{v_{\mathrm{rw}}{\left(t\right)}^2}{2}}\right)}^{{\displaystyle \frac{1}{2}}}{T}_{\mathrm{rw}}\left(1+c_{\mathrm{f}}\right)\hfill \\ & \left.+{\displaystyle \frac{m_{\mathrm{rw}}gv}{n_{\mathrm{r}}}}sin{\tau }_{\mathrm{c}}+{\displaystyle \frac{1}{2}}{d}_0{v}^3\rho c_{\mathrm{s}}A\right],\hfill \end{array}$$

(3)

where $\delta$ is the local blade section drag coefficient and A implies the disc area of each rotor. The thrust coefficient based on disc area and the rotor solidity are represented by $c_{\mathrm{T}}$ and $c_{\mathrm{s}}$, respectively, followed by $c_{\mathrm{f}}$ which represents the incremental correction factor of induced power. $\rho$ stands for the air density and ${\tau }_{\mathrm{c}}$ denotes the climbing angle of the UAV.

For clarity,

Table 2. Comparison of Two Types Commonly Employed UAVs.

Type of UAVs	Strengths	Weaknesses
Fixed-wing UAVs	Can manoeuvre at high speed and fly at great altitudes.	Likely to be affected by Doppler’s effect, it requires a runway or a catapult system to take off and has limitations on moving directions
Rotary-wing UAVs	Can manoeuvre in any direction or hover at a certain position.	High energy consumption and less efficient aerodynamics lead to limited flight times and ranges.

compares the difference between the two commonly exploited types of UAVs in IIoT applications.

4. Energy-Efficient Techniques in UAV-Assisted Wireless Communications

As discussed in the last section, employing UAVs in industrial wireless networks has various benefits and can significantly enhance the system performance. However, UAVs have stringent energy consumption constraints due to their limited size and weight. Hence, exploiting energy-efficient techniques is essential for the reliability of UAV-assisted networks. In this section, we present existing energy-efficient techniques in UAV-assisted wireless communication networks, including UAV placement, resource allocation, scheduling, beamforming and wireless power transfer. An example of energy-efficient UAV-assisted wireless networks is shown in Figure 3, where UAVs serve as relays and ABSs. We also observed that UAVs can either hover at a fixed position or fly along a trajectory while collecting data from users and devices, and they can support WPT within their coverage area.

4.1. UAV Placement

UAVs’ key characteristics of high flexibility and mobility set them apart from traditional TBSs. Optimizing the placement of UAVs is one of the most effective techniques for achieving energy-efficient wireless communication in UAV-assisted networks (e.g., in Figure 3 UAVs can hover or moving along the trajectory). There are two commonly adopted deployments of UAVs in existing works; the first method is to let the UAVs hover at a specific position during the operation and determine the optimal hovering locations [64,74,79,81,100,101], while the other option is to design the trajectories of the UAVs [70,72,75,76,78,93,102,103,104,105]. In this subsection, we introduce UAV placement techniques with corresponding adopted algorithms in various systems.

4.1.1. Optimal Hovering Position

Numerous research studies have focused on obtaining the horizontal coordinates and altitude of the UAVs to derive their optimal hovering location. Since UAVs are assumed to be fixed at a certain position in this scenario, only rotary-wing UAVs can be employed in the system. Also, if the UAVs’ position is fixed, their energy consumption remains constant since the speed and acceleration are all 0. Therefore, the energy consumption in the network is generally related to factors such as weather and wireless communication. Besides, the energy consumption of UAVs is interrelated to air density; it is important to note that UAVs operating at high altitudes increase their energy consumption because lower air density needs greater force to maintain hovering at a determined location [106].

According to [19], the two main factors related to weather that impact the energy consumption of the UAV are wind and temperature. If the UAV is in motion, wind can potentially aid in energy conservation depending on its direction. However, for hovering cases, UAVs have to consume energy to counteract the motion caused by the wind so that they stay in the hovering location. Compared to the wind, temperature affects the performance of the UAV’s battery since extreme temperatures lead to high battery drain. However, some of the UAV energy consumption papers that optimize the hovering location ignore the effect of weather (e.g., [64,101]).

Apart from considering weather factors, reducing the energy consumed by wireless communication is the common solution in existing papers to ensure energy efficiency. Considering single UAV operations, Ref. [64] optimized the transmission, computation offloading, and UAV deployment to maximize the system energy efficiency in a UAV-assisted blockchain operation IoT network. The authors separated the main problem into two layers where the outer layer for UAV hovering location was solved by a deep deterministic policy gradient (DDPG)-based learning algorithm, which obtains the optimal UAV deployment while enabling effective decision-making and adapting to the complex environment during the operation. Also, in [100], a UAV was deployed in a UAV-assisted WPT NOMA network. Again, three sub-problems were solved separately, where the UAV location sub-problem was solved by utilizing the Lagrange multiplier method. Furthermore, for multi-UAV cooperative networks, a coalition formation game-based solution was proposed in [79], where users function as players who form coalitions according to their utility values and then collaboratively secure the UAV’s service. Additionally, a k-means clustering algorithm was employed in [74] to obtain the optimal hovering location of the UAVs in a two-stage edge computing network while considering the mobile devices (MDs) association with the UAVs. Moreover, for multi-UAV-assisted NOMA networks, a maximum weighted independent set-based position optimization approach was adopted in [101] to address the reformulated problem by ignoring intercell interference and discretizing the deployment area while maintaining general applicability. By setting one UAV to be the relay, Ref. [81] determined the optimal hovering position of the UAVs to be the location that achieves maximum wireless coverage. Although obtaining optimal hovering locations can provide applicable results with less complexity and consumes less energy compared with trajectory planning, it cannot completely utilize the UAV’s high mobility since its position is fixed. Also, fixed-wing UAVs cannot be employed in these scenarios.

4.1.2. UAV Trajectory Design

To maximize the flexibility and mobility of UAVs, another line of research focuses on optimizing their trajectories to promote energy-efficient operation in UAV-assisted networks (e.g., [70,72,75,76,78,93,102,103,104,105,107,108]). Compared with the hovering position optimization scenario, most of the energy is consumed to maintain the motion. Although speed and acceleration both need to be considered according to the expression for UAVs’ energy consumption, most of the papers have neglected the impact of the change of acceleration on the energy consumption since the influence is tiny compared to the changes in speed. Also, it is worth noting that the energy consumption from other sources, such as wireless communication, is very low compared with flying energy consumption. Hence, some of the existing works have ignored even the energy consumption from other sources and focused on reducing flying energy consumption by designing trajectories for UAVs.

Considering single UAV scenarios, an optimization method that combines successive convex approximation (SCA) and a best-effort heuristic was proposed to tackle the energy efficiency problem in UAV-assisted blockchain-based IIoT networks [102]. The authors discretized the total operation duration into multiple short time slots and decided the flying speed in each of them. Also, in [103], the authors also adopted SCA for UAV trajectory planning for energy-efficient edge computing after obtaining results of other sub-problems and transformation of objective function and constraints. Again, SCA was exploited by [75] to find the sub-optimal UAV trajectory in a UAV-assisted NOMA–MEC network where the original nonconvex mixed-integer nonlinear fractional programming (MINLFP) problem has been transformed by the Dinkelbach method. Another UAV-assisted mobile edge computing paper [76] adopted a block successive upper-bound minimization (BSUM) algorithm, which continuously minimizes a sequence of tight upper bound of the objective function and sequentially updates the variables, including UAV’s location at each time slot. Additionally, Ref. [104] solved the energy-efficient problem in a UAV-assisted mobile edge computing by first exploiting SCA and Dinkelbach algorithm to transform the original non-convex problem and decomposed it by adopting an alternating direction method of multipliers (ADMM) technique to generate the UAV’s trajectory. By employing fixed-wing UAV, Ref. [93] optimized the UAV trajectory within the source and destination subspace while keeping the other fixed and utilized successive SCA and Dinkelbach methods to develop an algorithm that acquired a locally optimal solution for maximizing the energy efficiency of the UAV-assisted IoT network. Moreover, authors of [107] adopted a monotonic optimization framework for jointly optimizing the UAV trajectory and allocation of resources by transferring the original problem into the canonical form of a monotonic optimization problem. Although utilizing this approach can obtain the global optimal UAV trajectory and beamforming policy, the computation complexity increased rapidly with the number of users in the system. Also, it is worth noting that Ref. [107] took wind and no-fly zones of the UAV into consideration.

For multi-UAV scenarios, Ref. [78] decomposed the original energy-efficiency problem in a multi-UAV-assisted mobile edge computing network into three sub-problems and stated that the UAV trajectory planning sub-problem is actually convex with the result obtained by solving another sub-problem. Hence, the authors of [78] exploited $\mathtt{CVX}$ in $\mathtt{MATLAB}$ to obtain the optimal trajectories for the UAVs. To tackle the energy-efficiency issue in multi-UAV-assisted MEC networks, Ref. [70] separated the main problem and managed to reframe the non-convex constraints of the trajectory planning sub-problem, and employ an iteration-based SCA approach, where in each iteration, the main function is approximated by a simpler function at a specified local point. Considering multi-objective optimization (MOP), Ref. [105] formulated a complex hybrid optimization problem, including variables for the UAVs. To address this challenging problem, an improved discrete and continuous multi-objective evolutionary algorithm (EA) based on decomposition (IDCMOEA/D) has been developed to enhance the performance of initial solutions through a hybrid solution initialization method. Additionally, a hybrid solution reproduction operation was introduced to address coupled discrete and continuous problems, thereby improving the algorithm’s effectiveness in solving the formulated problem. Furthermore, Ref. [108] considered a multi-UAV system where one of them transmits data while the other UAV acts as a jammer to resist potential eavesdroppers. After decomposing the main problem and utilizing the results from the other subproblem, the optimal trajectory was obtained by exploiting $\mathtt{CVX}$ in $\mathtt{MATLAB}$ by going through a series of manipulations. Finally, authors of [72] utilized SCA to transform the constraints of UAVs’ trajectory sub-optimization problem and exploit $\mathtt{CVX}$ in $\mathtt{MATLAB}$ to obtain the optimal trajectories of the UAVs in a dual-NOMA-UAV-assisted IoT network. Compared with hovering location optimization, adopting the UAV trajectory planning technique leads to more complex computations and consumes more energy. However, it results in achieving a larger coverage area, addressing a wider range of tasks such as mapping, and being able to react according to actual situations. Hence, the overall energy efficiency can be improved, though UAVs have higher energy consumption than letting UAVs hover at fixed locations. Additionally, a brief summary of UAV placement techniques is provided in

Table 3. Comparison of Two UAV Placements.

UAV Placement	Articles	Strengths	Weaknesses
Optimal Hovering Position	[64,74,79,81,100,101]	Providing applicable results with less computational complexity and consumes less operational energy.	Cannot completely utilize the UAV’s high mobility or employ fixed-wing UAVs.
UAV Trajectory Design	[70,72,75,76,78,93,102,103,104,105,107,108]	Acquiring larger coverage area and addressing a wider range of tasks.	High computational complexity and energy consumption.

.

4.2. Resource Allocation

Another commonly adopted technique in modern UAV-assisted networks is resource allocation, since not only TBSs’ resources have been leveraged, but researchers also exploit UAVs as ABSs, as mentioned in the previous section and utilize their resources to improve the system performance while satisfying stringent requirements due to the significant increase of data and real-time decision-making process. Since UAVs and devices employed in the networks for wireless data transmission or WPT purposes (as shown in Figure 3) have limited resources such as power, time, bandwidth, channel, and offload computation, smartly distributing and utilizing them is essential for implementing the UAV-assisted networks in an energy-efficiency manner since factors such as the impact of imperfect CSI at transmitters or receivers. Additionally, existing papers often apply resource allocation jointly with other techniques, such as UAV placement, to further improve energy efficiency.

Power allocation for the transmit power of UAV or the devices is the most popular resource allocation technique adopted by various networks, including IoT [105], orthogonal frequency division multiple access (OFDMA) [108], NOMA [72,81] systems, UAV-assisted vehicular networks [109], ultra-dense networks [66] and for mMTC services [71], or edge computing [103]. Allocating more power to the channel with better performance is a promising method to enhance the system throughput and transmission rate while efficiently utilizing the energy at the device or UAVs. Also, power allocation is the most commonly adopted technique among all kinds of resource allocations and is often jointly optimized with other resources, such as time, computing resources and bandwidth.

Time allocation is commonly adopted in UAV-assisted networks with a time-division multiple-access (TDMA) protocol or scheduling. For hovering scenarios, time allocation normally focused on the transmission time duration of the UAVs and devices [110]. However, if UAV trajectory planning is considered, transmission time duration, flying time duration, or length of time slots must be optimized [69]. Also, it is worth noting that if UAVs are allowed to collect data during flight, time allocation for flying time is even more important [111].

Additionally, since the local computing resource of the server carried by the UAV is limited, offloading computing tasks to other devices, such as MEC servers, can overcome this issue and ensure energy-efficient operation [78]. For example, Ref. [112] optimized the offloading time, which predominantly influences the transmission time in the entire computation offloading process. Also, smartly allocating task computation and power resources can release the burden of processing large amounts of data across the network [75].

Apart from all these resources mentioned above, channel or bandwidth allocation is also exploited in UAV-assisted networks for an energy-efficient perspective. Utilizing techniques such as federated learning (FL) in bandwidth allocation occupies less bandwidth than other centralized methods by avoiding transmitting training data to a central server, significantly reducing communication expenses and network overhead [85]. Besides, a joint channel and power allocation technique adopted in NOMA was proposed in [101] for energy-saving purposes.

Moreover, allocating multiple kinds of resources and jointly optimizing them can capture various benefits related to corresponding variables, further enhancing performance and maximizing energy efficiency in UAV-assisted systems. According to the literature, existing works have achieved (1) joint power and time allocation [82,86,100,110], (2) joint power and computation resource allocation [75], (3) joint power and bandwidth allocation [65,79,113], (4) joint power and frequency allocation [70], (5) joint computation resource and bandwidth allocation [102], (6) joint power, bandwidth and computation resource allocation [74,76,85]. As we can see, utilizing power, bandwidth or computation resource allocation to enhance energy efficiency in UAV-assisted systems is the main trend in existing works. Also, adopting multiple resource allocation is a promising technique for achieving energy-efficient operation,

Table 4. Features of Resource Allocations in Existing Works.

Type of Resource	Articles	Related Features
Power allocation	[65,66,70,71,72,74,75,76,79,81,82,85,86,100,103,105,108,109,110,113]	Enhance the system throughput and transmission rate while efficiently utilizing the energy at the device or UAVs. The most common kind of resource allocation in existing works and is often jointly optimized with other resources.
Time allocation	[69,82,86,100,110,111]	Allocation for transmission time is mostly considered in UAV hovering scenarios while flying time duration and length of time slots also need to be optimized in trajectory planning conditions to achieve energy efficient implementation.
Computation resource allocation	[74,75,76,78,85,102,112]	Computational resource allocation is important to overcome the computing limitation issues of servers in the network, there are two approaches in the existing literature: (1) Directly allocating task computation; (2) Allocating the Offload process of the computing tasks to other devices.
Bandwidth/ Channel allocation	[65,74,76,79,85,101,113]	Conserving system bandwidth and reducing communication expenses and network overhead.

.

4.3. Scheduling

“Scheduling” is also one of the most popular energy-efficient techniques commonly applied in UAV-assisted networks. Unlike the similar technique of “time allocation”, which distributes specific amounts of time, scheduling focuses on planning and organizing tasks and events, considering prioritization and coordination among them. One method of exploiting scheduling is managing the network’s wireless communication and establishing transmission priorities. For instance, Ref. [73] implemented scheduling signalling between the UAV and IoT clusters so that only the cluster with the best conditions transmits without interference, thereby acquiring more data while consuming a certain amount of energy. Ref. [72] also adopted a similar scheduling technique to decide the connection between UAVs and NOMA groups to achieve better energy efficiency. Additionally, Ref. [114] controlled the system’s transmission time slot and destination of devices by exploiting scheduling parameters where a subcarrier-level scheduling approach was proposed in [108].

Scheduling techniques can also be applied to control the mechanism of devices in the network, in addition to scheduling communication. For example, in wireless sensor networks (WSNs), devices that keep monitoring environments or communicating might collect and transmit outdated data with energy consumption. Hence, the authors of [115,116,117] adopted a sleep-and-wake-up scheduling technique: the devices are allowed to stay in a low energy consumption “sleep” mode after completing the task or finishing sufficient mission progress. They turn to “wake-up” mode only in their scheduled time slot and communicate or start working at that time. This scheduling technique can effectively avoid unnecessary energy consumption and prolong the life span of UAV-assisted networks. Apart from scheduling for communication and devices mechanism, Ref. [102] adopted scheduling block generation at a UAV-assisted blockchain-based IIoT network. These existing works show that scheduling is a promising technique with great potential to cooperate with power or time-allocation-based techniques to further enhance their effects on energy-efficient improvements.

4.4. Beamforming

Considering UAVs’ high flexibility, directing their emitted energy in the intended direction is also a key technique for conserving energy and improving the throughput in UAV-assisted wireless communications [118]. Equipping UAVs with directional antennas can provide greater benefits than terrestrial channels by taking advantage of the high likelihood of LoS air-ground channels between UAVs and TBSs or users [17]. However, various technical issues such as serious interference between UAVs and channels can severely impact channel quality [119], leading to high energy consumption in wireless data transfer. To this end, employing the beamforming technique to ensure energy-efficient data transfer in UAV-assisted networks is promising. An example is [63], where the beamforming is achieved by exploiting the Deep Q-Network (DQN) method, which helps maximize energy efficiency with throughput requirements. Also, as mentioned before, Ref. [107] also employed beamforming factors optimizing techniques to enhance energy efficiency. Although literature seldom leverages beamforming for energy-efficient purposes, it has the potential to combine with other existing techniques, such as UAV placement and resource allocation. However, utilizing UAV trajectory planning together with beamforming techniques will result in high complexity in simulation due to the continuous changing of UAVs. Hence, it is important to investigate their trade-offs since servers and UAVs’ computation resources are limited in practical conditions. Another concern particularly encountered in UAV-assisted massive multi-in-multi-out (M-MIMO) systems is the accuracy and availability of channel state information (CSI) at TBSs and users since it significantly affects the beamforming gain [120]. Below in

Table 5. Features and Potential of Two Techniques in Existing Literature.

Techniques	Articles	Related Features	Potential
Scheduling	[72,73,102,108,114,115,116,117]	Focuses on planning and organizing tasks and events, including exploiting mechanism of devices, control the connection and prioritization of the users.	Can cooperate with power or time allocation-based techniques to enhance energy-efficient implementation.
Beamforming	[63]	Fully utilizing the LoS air-ground channels provided by UAV-assisted communication while mitigating the serious interference in UAV-assisted communications.	Can be adopted with other existing techniques, such as UAV placement and resource allocation. It is worth noting that utilizing beamforming in the UAV trajectory planning scenario is very challenging.

we briefly summarize the features and potential of “scheduling” and Beamforming techniques.

4.5. Wireless Power Transfer

Furthermore, the UAVs and devices deployed in UAV-assisted wireless networks are often powered by batteries. Due to the energy storage limitations of these batteries, they have to be charged or replaced periodically. However, in practical IIoT applications, such as WSNs, sensors are distributed in a large area, and some of them are hard to reach. In these scenarios, replacing batteries can be challenging and expensive and consume considerable human and financial resources. Overcharging the batteries brings potential risks such as leakage, which can destroy and pollute the environment [121]. As a remedy, WPT techniques are utilized in UAV-assisted networks to enhance energy efficiency, ensure sustainable system implementation, and prolong the lifetime of the network. One popular WPT technique exploited in the existing literature was to let UAVs harvest energy from other sources. For instance, Ref. [100] considered an energy-efficient UAV-assisted NOMA network in which a power beacon was established to transmit energy wirelessly to the UAV. Another option for UAVs is to harvest energy from the access point as proposed in [112] to maximize energy efficiency while prolonging UAVs’ service time. However, the construction cost of infrastructures is costly, and their wireless service coverage is often limited [122].

Instead of harvesting energy from other devices and facilities, UAVs can also serve as power sources, employing WPT to charge other devices as shown in Figure 3. For instance, Ref. [82] achieved energy efficiency UAV-assisted network with WPT from UAVs to the user devices deployed in the system. By taking public safety into account, Ref. [84] utilized UAVs to charge the IoT devices in a UAV-assisted NOMA network wirelessly. By enabling rechargeable batteries on user devices, Ref. [85] allowed them to harvest energy from the UAV’s radio-frequency signals and exploit environmental energy such as solar power. Also, Ref. [83] proposed a framework where the UAV can transmit data and harvest energy from TBSs, processing its tasks and charging the IoT nodes while collecting data from them. Furthermore, not only do UAVs and devices transmit power to each other, but WPT techniques can be leveraged in machine-to-machine (M2M) communication and support to enhance energy efficiency [86]. Although adopting WPT can effectively utilizing the energy in the system while ensuring energy efficiency, due to the limited battery storage of UAVs and devices in the network, guaranteeing sufficient energy for WPT is the main challenge for utilizing this technique,

Table 6. Two kinds of Wireless Power Transfer and Related Challenges.

Wireless Power Transfer	Articles	Strengths	Weaknesses
From infrastructures to UAVs	[100,112]	Ensuring UAVs have sufficient energy for completing their missions and ensuring system reliability.	Infrastructures like power beacons have high construction costs and limited wireless service coverage.
From UAVs to devices	[82,83,84,85,86]	Prolonging network work time by performing sustainable operation and preventing potential risks such as battery leakage of devices.	UAVs have stringent constraints on size and weight which leads to concerns on energy storage for achieving WPT on a continuous basis.

.

5. Energy-Efficient Techniques for UAV-Assisted Industrial Wireless Networks

Building upon the comprehensive review of energy-efficient techniques in industrial networks, UAV capabilities in industrial wireless networks, and energy-efficient methods in UAV-assisted wireless communications presented in the preceding sections, this section aims to examine the existing energy-efficient techniques specifically tailored for UAV-assisted industrial wireless networks. Comparing with common UAV-assisted wireless communication systems, UAV-assisted industrial wireless networks’ architectures are specifically designed for targeted missions [123] and are integrated with industrial protocols and systems [124], as shown in Figure 4. Additionally, reliability, security, and safety standards gather more significant concerns in UAV-assisted industrial wireless networks’ operation process [125].

Drones have garnered considerable attention due to their high capability and energy-efficient characteristics, positioning them as a critical component for next-generation communication systems. Hence, this section will be divided into two subsections. The first one details the application of energy-efficient techniques in UAV-assisted industrial wireless networks, elucidating their operational mechanisms and the necessity of integrating UAVs. Subsequently, we will identify several open challenges in this emerging field.

5.1. Existing Energy-Efficient Techniques

In this subsection, we analyze existing energy-efficient techniques, highlighting their necessity and how they contribute to enhancing energy efficiency within the framework of UAV-assisted industrial wireless networks. We will also explore why the integration of UAVs is essential and how it benefits the framework. This discussion aims to provide a deeper understanding of both UAV and industrial wireless networks. Below Table 7 provides a concise summary of the existing energy-efficient techniques along with their respective strengths and concerns within the context of the IIoT.

5.1.1. Vehicular Networks and Platooning Systems

The industrial sector has reaped significant benefits from vehicular networks, which enhance road capacity and fuel efficiency, and contribute to supply chain improvements by reducing human labour costs. However, vehicular networks often operate in unpredictable communication environments, which can complicate the requirements for multi-vehicle cooperative control. To overcome these challenges, the study by Duan et al. explore a UAV-assisted multi-vehicle cooperative platoon system designed to enhance the communication success probability between vehicles and UAV nodes while reducing velocity perturbations during transit to support the energy-efficient transportation [126]. Beyond merely acting as communication relays, UAVs have demonstrated potential as sensors, as noted in [127], providing critical information for improved vehicle trajectory planning and, therefore, reducing fuel consumption by avoiding sudden velocity changes.

Additionally, with the rapidly increasing demands for pervasive communication and computing, terrestrial vehicular networks alone are insufficient to satisfy these needs effectively. Integrating MEC facilities, such as cloudlets, onto UAVs can provide edge computing services from the sky, addressing the resource constraints faced by terrestrial networks [128]. To further enhance the energy efficiency of UAV-assisted vehicular platooning systems, the adoption of wireless power transmission has been proposed as a means to extend the service duration of UAVs, thereby improving overall system energy efficient performance [129].

5.1.2. UAV-Assisted Backscatter Communication

As discussed in Section 2, backscatter communication is recognized as a green and energy-efficient communication technique, showing great potential for realizing future IoT networks. It facilitates connectivity among a vast array of smart devices across diverse applications such as industrial automation and agricultural industrialization [130,131]. Simultaneously, the high mobility of UAVs allows them to serve as effective information collectors. Additionally, UAVs can support both wireless power transfer and communication platforms, thereby enhancing energy efficiency [132]. Their dual functionality, combined with the low power requirements of backscatter networks, renders them particularly suitable for advancing agricultural industrialization. Moreover, the work in [133] further optimized the energy efficiency of UAV-assisted backscatter communication by improving UAV data collection locations. Additionally, a joint trajectory and resource optimization strategy proposed in [134] aimed to enhance energy efficiency, bringing these systems a step closer to practical implementation.

5.1.3. RIS-Assisted UAV Systems

As previously noted, RIS are lauded not only for their low energy requirements but also for their ability to enhance wireless communication by serving as relays. In 2022, Liu et al. proposed a secure RIS-assisted UAV system in a complex IIoT environment. This system, which serves multiple ground users, aims to maximize the minimum average security rate using an efficient algorithm based on the block coordinate descent (BCD) algorithm [135]. Further advancing energy efficiency, Qin et al. a year later introduced a joint optimization of resource allocation, phase shift, and UAV trajectory for RIS-assisted UAV-enabled MEC systems, employing the low-computation-cost BCD technique [136].

Additionally, RIS-assisted communication frequently integrates other energy-efficient techniques, such as WPT. Specifically, IIoT devices harvest energy from UAVs through wireless power transfer, after which the UAV collects data from these devices via information transmission [137]. In this scenario, the power and trajectory of the UAV, along with the scheduling of the devices, are jointly optimized exploiting deep reinforcement learning, requiring minimal processing time. Addressing the demands of Industry 4.0, which requires the capability to connect massive numbers of IoT devices reliably and seamlessly anywhere and at any time within the manufacturing industry, Xu et al. explored the use of UAVs and RIS to provide robust air-to-ground links. They also introduced D2D communication techniques to facilitate direct information exchange between IoT devices. The transmit power, channel allocation parameters, and RIS’s reflection coefficients are jointly optimized to maximize energy efficiency for D2D users, employing both centralized and distributed optimization algorithms based on deep neural networks (DNN) [138].

5.1.4. Machine Learning-Based Approaches

In addition to specific energy-efficient techniques, various other approaches significantly enhance UAV-assisted industrial wireless networks. In 2021, shi et al. employed MEC to handle tasks requiring high QoS for energy-constrained IIoT devices. They utilized UAVs equipped with transceivers as aerial MEC servers, offering IIoT devices opportunities for computation offloading to increase server flexibility. Furthermore, they introduced an intelligent computation offloading algorithm termed multi-agent deep Q-Learning with stochastic prioritized replay (MDSPR), which achieves an energy-efficient and low-complexity solution and demonstrates rapid convergence and robust performance [62]. Moreover, considering that machine-type communications devices (MTCDs) are typically battery-powered, limited energy storage remains a significant concern. The same year, Yan et al. proposed a UAV-assisted network equipped with remote MTCDs to provide new value-added services to maximize energy efficiency within machine-type communications [139]. Additionally, the industrial internet of unmanned aerial vehicles (IIoUAVs), which facilitate autonomous inspection and measurement capabilities accessible anytime and from anywhere, have become a crucial component of the future IIoT ecosystem. Specifically, the application of IIoUAVs for power line inspection within smart grids, considered from an energy efficiency perspective, seeks to minimize energy consumption and enhance operational efficacy [140].

Table 7. Strengths and Concerns of Existing Energy-Efficient Techniques in IIoT.

Techniques	Articles	Strengths	Concerns
Vehicular networks	[126,127,128,129]	Enhancing fuel consumption reduction along with improving road capacity and safety to benefit supply chains.	Significant Doppler shift challenges; reliability and safety of communication are difficult to guarantee.
Backscattering	[130,131,132,133,134]	Low power consumption and low cost; each device has its own ID for easy tracking.	Communication range is limited.
Reconfigurable intelligent surfaces	[135,136,137,138]	Low power consumption; enhances communication performance with extended range.	Highly dependent on propagation environment; and the RIS-related channels are hard to estimate and control.
Machine learning-based approaches	[139,140]	Easy to tackle non-convex or challenging-formulated problems.	High costs; demands substantial computational resource demands and large datasets.

Table 7. Strengths and Concerns of Existing Energy-Efficient Techniques in IIoT.

Techniques	Articles	Strengths	Concerns
Vehicular networks	[126,127,128,129]	Enhancing fuel consumption reduction along with improving road capacity and safety to benefit supply chains.	Significant Doppler shift challenges; reliability and safety of communication are difficult to guarantee.
Backscattering	[130,131,132,133,134]	Low power consumption and low cost; each device has its own ID for easy tracking.	Communication range is limited.
Reconfigurable intelligent surfaces	[135,136,137,138]	Low power consumption; enhances communication performance with extended range.	Highly dependent on propagation environment; and the RIS-related channels are hard to estimate and control.
Machine learning-based approaches	[139,140]	Easy to tackle non-convex or challenging-formulated problems.	High costs; demands substantial computational resource demands and large datasets.

5.2. Open Research Problems, Challenges and Future Directions

In this subsection, we highlight open research challenges based on the literature reviewed and analyzed in previous sections, including adopting accurate models, implementing practical prototypes, harnessing renewable energy and ensuring privacy and security.

5.2.1. Limitations of Existing UAV Communication Models

Based on our discussion and analysis of the literature, the first challenge in existing UAV-assisted industrial wireless networks is the limitations of current UAV communication models. Recently, researchers have devoted themselves to investigating energy-efficient techniques in theoretical conditions. However, reliability remains a stringent requirement under industrial wireless networks, where analytical results from theoretical papers cannot be completely guaranteed in actual cases. Also, most of the existing works have adopted strong assumptions such as a fixed UAV altitude during flight, perfect CSI at transmitter and receiver, ignoring weather’s effect and exploiting theoretical models. For instance, due to various practical factors, maintaining a constant UAV altitude throughout an operation is challenging. Moreover, its 3D movement capabilities cannot be fully exploited if the UAV remains at a fixed altitude. Additionally, CSI is an important factor in wireless communications. In trajectory planning scenarios, estimation errors due to variations in UAVs’ position and finite data feedback [141] can degrade the accuracy of CSI significantly. Hence, considering imperfect CSI would improve the practicality of the designs in real-world operations. Another factor that has often been overlooked is the weather. As mentioned before, UAV’s energy consumption is related to wind, especially in hovering location optimization conditions. Additionally, the influence of temperature needs to be considered, since fixed-wing UAVs operate at high altitudes, where the temperatures are lower, and extreme temperatures can reduce the batteries’ performance.

Besides reducing reliance on these assumptions, discovering and adopting accurate models are also essential to obtain more practical results. For example, existing UAV-assisted wireless energy harvesting papers adopt linear energy harvest models, but leveraging non-linear wireless energy transfer models can obtain more practical results. Also, UAV’s available energy needs to be considered since WPT imposes stringent constraints on the energy level of the transmission source [142]. Another commonly adopted model in multi-UAV-assisted networks is the air-to-air (A2A) channel model. Most of the authors utilized the Rician model for A2A communication links but might not be able to adopt it in high sampling rate conditions [143]. Hence, conducting experiments that leverage theoretical methodologies and corresponding techniques to obtain more practical results and implementing experiments to find better models is one of the future works. More accurate UAV communication models, such as the A2A channel model, can be developed by analysing results obtained from actual scenarios. Subsequently, to achieve better analytical results, researchers should reduce reliance on assumptions (e.g., fixed UAV altitude, perfect CSI and sufficient energy for UAVs to complete all missions) while exploiting accurate models from practical analysis in future studies.

5.2.2. Implementing Practical Prototypes

Although numerous studies have investigated both UAV systems and industrial wireless networks, as well as their joint framework, the majority of this research has focused on mathematical simulations rather than field testing. This predominance of theoretical over practical validation may lead to inaccurate results and conclusions, primarily due to the high mobility inherent in UAV and vehicular communication systems and the challenges in predicting rapidly changing environmental conditions. For example, widely used vehicular communication standards such as IEEE 802.11 bd and ETSI [144], which operate around the 5.9 GHz frequency range, can experience significant Doppler shifts due to the high mobility of vehicles [145]. However, only a few studies have taken this factor into consideration. Moreover, the complex urban environment also significantly affects path loss and scattering, frequently alternating between LoS and Non-Line-of-Sight (NLoS) conditions. The variability and unpredictability of these conditions render real-time resource optimization practically infeasible due to the high computational costs driven by rapidly changing channel conditions.

Additionally, although numerous researchers have endeavoured to advance and validate energy-efficient technologies, there is a notable lack of practical prototype implementations to verify their real-world functionality. While there are some groups that have begun field testing backscatter systems in mobile scenarios [146], the consistency and reliability of these tests still require further verification. Especially in industrial networks, where the deployment of massive device arrays is necessary, practical prototype testing is crucial before progressing to more comprehensive implementation stages. Consequently, the future work is to fulfil the necessity for practical applicability of theoretical models through robust field testing, and prototype development, which is imperative for advancing UAV-assisted industrial wireless networks towards real-world implementation, particularly for energy-efficient ITS.

5.2.3. Utilizing Renewable Energy Sources

To address the energy shortage issue in UAV-assisted industrial wireless networks while ensuring the batteries of the devices in the network have sufficient energy to support them in finishing their tasks, existing papers have considered options such as replacing the devices’ batteries or exploiting WPT to recharge them. Also, some of the UAVs employed in existing UAV-assisted industrial networks are fossil fuel-powered, which is not eco-friendly due to their greenhouse gases emission [147]. Besides, electrical-powered UAVs have very limited endurance [148]. Hence, utilizing renewable energy for UAVs and devices deployed in the network is practicable. One renewable energy that can be exploited is solar energy [149]. By equipping solar panels on the UAV [150] or devices, they can harvest energy themselves and release the burden of batteries and WPT from other sources. Additionally, employing high-altitude platforms (HAPs) equipped with solar panels and wireless charging capabilities for UAVs has the potential to enhance energy efficiency, as the UAVs would not need to climb up to seek more solar power. Exploiting wind energy in windy areas to keep the UAV airborne for extended periods is also a viable method for achieving better energy efficiency. Utilizing the kinetic energy of wind to generate power for UAVs and devices can be a potential solution to the energy shortage in actual UAV-assisted IIoT applications. From a fuel perspective, leveraging hydrogen fuel cells to generate electricity through a chemical reaction can increase energy efficiency and be eco-friendly [151] since only water vapour is emitted. However, producing hydrogen and building the infrastructure for refuelling the cells will be challenging. Hence, utilizing renewable energy sources, such as wind and solar power, to extend the energy storage of devices and UAVs not only sustains the network but also ensures eco-friendly operation in the future research.

Another area of future work involves the extension adoption of WPT techniques in these systems. By facilitating WPT between UAVs, devices, and infrastructures via power beacons, energy levels can be maintained to accomplish missions, perform real-time decisions, or handle potential accidents. Additionally, leveraging resource allocation and scheduling techniques can optimize the utilization of power, time, and channels in the wireless communication layer, thereby efficiently organizing tasks to support WPT. For example, by allocating power across channels or subcarriers and transmitting in specific time slots, data can be gathered efficiently with low energy consumption. Furthermore, adopting scheduling strategies such as the “sleep-and-wake-up” approach can prevent energy wastage due to overcharging or unnecessary WPT from power sources.

5.2.4. Ensuring Privacy and Security

Furthermore, concerns about privacy and security are crucial in UAV-assisted industrial wireless networks due to the sensitive nature of data transmitted and the potential risks associated with unauthorized access and eavesdroppers. UAVs can provide strong LoS communication links at low cost, but information signals sent over wireless LoS channels are generally susceptible to interception by unauthorized receivers [152], increasing the risk of information leakage. Additionally, wireless UAV transceivers are prone to malicious jamming attacks [153]. However, numerous existing works only focus on achieving maximum energy efficiency but overlook the importance of network security. Hence, in the future reseach, utilizing safety protocols and leverage data encryption can be one of the options for improving the security of sensitive data. Ref. [154] proposed a drone security module to safeguard UAVs’ communication data and stored information, including the module’s hardware and software driver designs, which connect to the flight control or mission computer via a universal serial bus (USB). Additionally, adopting security protocols can also assist UAV-assisted wireless communication, and there are existing safety communication protocols such as UranusLink protocol, UAVCAN protocol and MAVLink protocol [155]. Also, utilizing routing protocols such as a Model Driven Development (MDD) approach [156] or encrypts routing messages between UAVs and ground stations using digital signatures and an asymmetric encryption mechanism [157].

Besides, as the number of connected devices grows, ensuring data transmission security becomes a prerequisite for adopting any of the aforementioned energy-efficient techniques. This is particularly crucial when considering low-power or passive technologies, such as RIS and backscatter communication. These technologies are constrained by limited computational resources and size, making data protection more complex. Therefore, future investigations on physical layer security gather great potential for addressing data transmission concerns within wireless industrial networks. The machine learning-based approach is also a popular way to optimize the IIoT performance [158], especially bringing a performance improvement in the recognition rate of unknown malicious attacks [159]. Moreover, exploring techniques is essential to ensure privacy and security in UAV-assisted industrial wireless networks while satisfying energy-efficient constraints. Additionally, to further optimize the energy efficiency of UAV networks, several studies have implemented green backscatter techniques [38]. It should also be noted that backscatter tags are constrained by their size and computational resources. On the other hand, although physical layer security is an appropriate technique for securing transmissions in dynamic mobility scenarios, its unique characteristics also challenge the channel model [160]. Subsequently, the computational cost is a limitation for adopting it in the real world. Furthermore, secure UAV operation can also be achieved by optimizing the trajectory and resource allocation [153]. Hence, the beamforming techniques and resource allocation at the transmitter represent a more feasible approach than directly optimizing the parameters within the tag. Moreover, leveraging UAVs as rotating or mobile antennas [161] and transmit on specific directions during allocated timeslots is a potential solution for increasing transmission security. Another possible solution is utilizing one or a set of UAVs serving as jammers to resist eavesdroppers or even utilizing eavesdroppers’ signals for charging the devices and UAVs in the network. Consequently, one of the future directions is adopting beamforming techniques together with the flexibility of the UAVs while considering resource allocation and exploiting “scheduling” to smartly adjust their roles and missions to enhance security while reducing energy consumption.

6. Conclusions

This paper aims to provide a comprehensive review of energy-efficient techniques in UAV-assisted industrial wireless networks. First, we introduced the background of IIoT and UAV-assisted networks with existing reviews and motivation. Next, we analyzed and discussed various energy-efficient techniques together with related papers. To overcome stringent limitations in traditional industrial wireless networks, we highlight UAV’s capability and potential to assist IIoT applications. However, employing UAVs consumes even more energy, so energy-efficient techniques are essential for reliable system operation. To this end, we reviewed existing energy-efficient techniques with related papers and algorithms, with a focus on UAV placement. Finally, we identified several energy-efficient techniques for UAV-assisted industrial networks to highlight open research problems, corresponding recommended approaches, and possible improvements. Furthermore, we discuss and summarize existing challenges and future directions in the research area. We hope this paper will serve as an inspiration and guide for the design and development of energy-efficient techniques in UAV-assisted industrial wireless networks in the future. Below, we highlight and summarize the fourfold contribution of this paper again:

(1) Provided detailed reviews of energy-efficient techniques in industrial wireless networks and UAV-assisted networks and analyzed their features, strengths, weaknesses, and potentials.

(2) Identified UAV’s capabilities, including their roles as aerial base stations (ABSs), network expansion, and cost reduction. We explained the reason for leveraging UAVs in traditional industrial wireless networks and provided energy consumption models for two types of commonly employed UAVs.

(3) Highlighted existing energy-efficient UAV-assisted industrial wireless networks and discussed the energy-efficient techniques in this literature.

(4) Articulated open research challenges in UAV-assisted industrial wireless networks, such as developing accurate models and utilizing renewable energy sources. Additionally, we have outlined future research directions.