Extremum Seeking-Based Radio Signal Strength Optimization Algorithm for Hoverable UAV Path Planning

Abstract

For the safe autonomous operations of unmanned aerial vehicles (UAVs) and ground control stations (GCS), including autonomous battery replacement, wireless power transfer, and more, the precise landing of UAVs on GCS is essential. Accurate landing is only possible when the link capacity strength exceeds a certain threshold, but this is often disturbed due to complex terrain. To address this, we developed an extremum seeking (ES)-based radio signal strength optimization (RSSO) algorithm, ES-RSSO, designed to find the optimal positions of the UAV using radio communication signals. This ensures energy-efficient path planning while guaranteeing the minimum received signal strength indication (RSSI) capacity. This algorithm is particularly useful in obstacle-rich environments, where UAVs are limited in power resources. Simulation results demonstrate a 2.37% decrease in the mean, a 62.08% improvement in variance, and a 3.72% decrease in the integration strength of the link capacity when ES-RSSO is applied. These results confirm that the RADIO.rssi maintenance ability remains above a critical boundary level, supporting robust communication links and energy-efficient path planning. Throughout the study, we showed how, in many cases, simply moving the UAV a few meters can significantly improve the communication link.

1. Introduction

1.1. Background

An uncountable number of small and lightweight unmanned aerial vehicle (UAV) applications have emerged over the past decades, and crashes involving UAVs, leading to hazardous accidents, have also increased. These accidents can be caused by various factors, including improper handling, battery usage beyond safe limits, carelessness, strong winds, detached propellers due to loose screws, communication disconnections, or sensor malfunctions such as global positioning system (GPS) or inertial measurement unit (IMU) failures due to strong magnetic fields. Among the causes of UAV accidents listed above, it was preliminarily analyzed and revealed that nearly one-third of the cases examined experienced a loss of communication link [1].

We propose an extremum seeking (ES) [2,3,4]-based radio signal strength optimization (RSSO) algorithm, ES-RSSO, to enhance the wireless data link capacity between the UAV and the ground control station (GCS). This ensures strong and stabilized communication while achieving energy-efficient path planning and guaranteeing a minimum received signal strength indication (RSSI) capacity.

The RSSI in the radio communication between the UAV and the GCS represents a power measurement at the receiver [5]. Although the definition of RSSI is outlined in the IEEE 802.11 standard [6], the unit of RSSI is not explicitly defined in the standard. Instead, the power level measured at the receiver is conventionally expressed in milliwatts ($\mathrm{m}\mathrm{W}$), or sometimes in decibels relative to one milliwatt ($\mathrm{d}\mathrm{B}\mathrm{m}$), where 0 $\mathrm{d}\mathrm{B}\mathrm{m}$ equals 1 $\mathrm{m}\mathrm{W}$ [7].

The RSSI typically decreases when the UAV is far from the GCS or blocked by obstacles and increases when it is close to the GCS [8,9]. In UAVs equipped with Pixhawk [10] and Mission Planner [11], the micro air vehicle link (MAVLink) [12] support in SiK radios [13] includes two key message parameters: RADIO.rssi and RADIO.remrssi [14]. RADIO.rssi represents the RSSI level received by the local radio, while RADIO.remrssi represents the RSSI level received by the remote radio.

This paper focuses on maintaining reliable UAV communication by ensuring the signal strength (RADIO.rssi) stays above a threshold, enabling energy-efficient path planning. We show that slight UAV movements can greatly enhance the communication link, offering an energy-saving approach. Using the ES-RSSO algorithm, strong communication with minimal power supports mission plans like autonomous wireless charging [15,16,17].

1.2. Literature Review

The literature review of research related to the proposed ES-RSSO-based path planning is summarized in

Table 1. Comparative analysis of the proposed method with the related literature.

	Path Optimization		Distance Extension Optimization	Coverage Optimization
Literature	Propose Method	[18,19,20,21,22,23,24,25]	[26,27,28]	[29,30,31,32,33,34]
Basic Ideology	Radio map-based path planning	Xest path planning (shortest, fastest, etc.)	Communication beyond the limits of telemetry range	Coverage path planning (CPP)
Mission Goal	- Mission time - Energy efficiency	- Mission time - Energy efficiency	Multi-UAV-based communication relay	Resource allocation
Optimization Technique	ES-RSSO	- Graph theory-based shortest path problem (SPP) - Successive convex approximation (SCA) - Rapidly-exploring random trees (RRT) - Quantum-inspired experience replay (QiER)-based deep reinforcement learning (DRL) - Inverse reinforcement learning (IRL)-based Q-learning and DRL	Heuristic convex optimization	- Particle swarm optimization algorithm based on decomposition (PSO/D) - SCA and block coordinate descent techniques - Mixed integer linear programming (MILP)-based unit disk graph (UDG) model - Path loss model and decision tree machine learning classifier
Pros	- Fast convergence time to maximize or minimize unknown dynamic functions - Widely verified robustness [35]	Better Xest path optimization achieved through the Q-learning algorithm based on the Markov decision process and DRL [36]	A relatively simple yet efficient iterative algorithm	- Simple framework - High convergence speed
Cons	Semi-global convergence	- Relatively slow computation time for optimum convergence - High training time and data requirements for DRL

, with detailed analysis in the following subsections. This method is more advantageous for practical applications due to its fast convergence and suitability for real-time path planning.

1.2.1. Path Optimization

If the UAV operates within the range of the telemetry modem and a robust communication strength is required from departure to destination, a path optimization method will be needed to maintain communication quality.

Zhang et al. developed a methodology to obtain the radio map-based optimal 3D path based on an equivalent SPP in graph theory [18]. Dong et al. presented a UAV anti-jamming communication methodology by incorporating two radio maps (signal power and signal-to-interference plus noise ratio (SINR) maps) into the heuristic, graph-based A* algorithm to achieve the optimal path based on the SPP [19]. Zhou et al. developed a non-convex semi-infinite optimization formulation to maximize the average worst-case SINR under strict energy availability constraints, using the S-procedure and SCA methods [20]. Khamidehi et al. developed a two-step path optimization method; (1) federated learning-based outage probability model generation, and (2) RRT-based path planning [21]. El-Gayar et al. proposed a 6G network model and caching strategy using UAVs, based on non-orthogonal multiple access and orthogonal multiple access systems, to improve the cache hit rate over time. This ultimately results in improvements in energy efficiency, content distribution, latency, and transmission speeds in communication [22]. Zhang et al. presented convex optimization and graph theory techniques to achieve a near-optimal solution for minimizing UAV mission completion time, with an optimized path subject to the minimum signal-to-noise ratio (SNR) requirement and the maximum UAV speed constraint [23]. Li et al. proposed a QiER-based DRL technique to simultaneously minimize the time cost and expected outage duration, aiming to establish a high-quality ground-to-air (G2A) transmission-capable UAV path [24]. Shamsoshoara et al. utilized IRL-based Q-learning and DRL to maximize the G2A uplink throughput and minimize ground user equipment interference [25].

1.2.2. Distance Extension Optimization

If the UAV flies beyond the range of the telemetry modem, the level of RADIO.rssi drops below a critical threshold, eventually leading to a loss of communication with the GCS. Numerous studies have been conducted to increase communication distance.

Li et al. employed a communication relay system between two UAVs to extend the communication range and overcome beyond-line-of-sight (BLOS) situations [26]. Wu et al. extended the range of BLOS UAV communication by utilizing multi-UAV mounted aerial flying base stations (FBS) and optimizing communication scheduling, association, trajectory, and power control [27]. Tran et al. addressed the UAV relay-assisted internet of things (IoT) wireless network by employing full-duplex radio communication to overcome both the data transfer latency of IoT sensors and the data storage limitation of UAVs, focusing on resource allocation and trajectory optimization [28].

Compared to the references listed above which utilize the concept of the UAV relay to extend communication distance, the proposed ES-RSSO algorithm can find energy-efficient paths (i.e., requiring less energy consumption) in normal situations and extend the communication distance in emergency situations by increasing the signal strength using leftover energy.

1.2.3. Coverage Optimization

If UAVs fly over a boundary region within a flying ad hoc network (FANET) while minimizing time or energy consumption, a CPP optimization technique is required.

Daniel et al. proposed a method leveraging public wireless communication infrastructures, utilizing civilian concepts of operations to address chemical, biological, radiological, and nuclear reconnaissance. This approach aims to overcome communication delays and reliability constraints [29]. Pan et al. presented a two-stage approach using pre-programmed locations and PSO/D for UAV trajectory and resource allocation, aiming to maximize average throughput under the constraints of co-channel interference and completion time [30]. Shi et al. proposed a UAV CPP method that satisfies the required network quality of service (QoS) by utilizing a heuristic PSO algorithm [31]. Hu et al. developed SCA and block coordinate descent techniques to optimize UAV paths over large areas, minimizing propulsion energy while achieving the sensing resolutions required for communication-assisted radar sensing [32]. Lee et al. presented a MILP-based UDG model to optimize multiple UAV flight times for the FBS missions, aiming to satisfy QoS while considering 5G/6G mobile SNR wireless network interference [33]. Khan et al. introduced a multiple UAV-based FANET strategy that considers RSSI to identify the location of wireless nodes in both 2D and 3D by developing a path loss model and a decision tree machine learning classifier [34].

1.3. Contribution

To maintain communication signal strength above a preset boundary for energy-efficient and minimum RSSI capacity-guaranteed path planning, we propose a method to find the local minimum RSSI value as the UAV flies along the planned trajectory using the ES-RSSO algorithm. This method is particularly useful when the UAV is located in relatively complex environments. It can be applied by iteratively running the ES-RSSO algorithm, simply reinitializing the UAV’s position to one different from the previously visited local maximum. The ES-RSSO algorithm computes the optimized RSSI path by identifying the gradient of the link capacity map, generated from the RSSI data using a washout filter, demodulator, integrator, and additional components.

1.4. Article Organization

The structure of this paper is as follows. Section 2 introduces the mission. Section 3 outlines the problem formulation, and Section 4 describes the simulation results. Lastly, Section 5 presents the conclusion and discusses future work.

2. Mission

2.1. Scenario

The UAV is equipped with sensors, a flight controller (FC), an optimization module, and a transceiver, as shown in Figure 1. The UAV flies from a takeoff waypoint to a landing waypoint while collecting RSSI data in real-time, computing to find the local minimum, and finally relocating to the identified local minima.

2.2. Typical RSSI and MAVLink Log Graphs

Figure 2a shows a typical graph of the RSSI levels for the radio communication between the UAV and the GCS. As depicted in Figure 2a, the signal strength clearly decreases when the UAV moves farther from the GCS and increases as it approaches the GCS. MAVLink support in SiK radios includes two key message parameters: RADIO.rssi and RADIO.remrssi. The RADIO.rssi represents the RSSI level received by the local radio, while the RADIO.remrssi represents the RSSI level received by the remote radio.

Figure 2b shows a typical MAVLink log graph of the radio communication between the UAV and the GCS. The four graphs in Figure 2b represent the signal strength received by the GCS, the signal strength received by the UAV, the noise received by the GCS, and the noise received by the UAV.

2.3. Overview of the ES-RSSO Algorithm

Figure 3 illustrates the overall logic of the ES-RSSO algorithm, which is used to determine the position of the minimum link capacity in the radio communication between the UAV and the GCS. The GCS, running MATLAB R2024a software with the ES-RSSO algorithm, receives the Pth UAV position ${\Theta }_p$ from the physical UAV through the MatMav communication interface [37]. The ES-RSSO algorithm then computes the optimal Pth UAV position ${\Theta }_p^{*}$, which has the lowest RSSI link capacity, and uses it as the next waypoint of the Pth UAV.

3. Problem Formulation

3.1. Problem Setup

The UAV continuously seeks the local minimum position where the link capacity reaches its lowest local value. Additionally, the UAV must fly from the takeoff waypoint to the landing waypoint in the shortest possible time. This fundamental problem can be formulated as follows:

$$\mathit{min}E\left[\sum r\left({\Theta }^{*}\right)\right],\hspace{1em}s.t. \mathit{min}E\left[T\right],$$

(1)

where $\sum r$ represents the sum of the RSSI link capacity during the flight from the departure to the destination location ($\mathrm{d}\mathrm{B}\mathrm{m}$), ${\Theta }^{*}$ is the optimal value of $\Theta$ ($\mathrm{m}$), $\Theta$ is the position coordinate of the UAV ($\mathrm{m}$), i.e., $\Theta =\left[x,y,z\right]$, and $T$ is the mission completion time ($\mathrm{s}$).

To solve Equation (1), the hierarchy of mission planning is constructed as shown in Figure 4, which outlines the detailed steps described in Figure 3. According to Figure 4, the UAV continuously measures the RSSI link capacity, replacing the existing lowest RSSI value with any newly detected higher RSSI value. Since the ES-RSSO algorithm cannot always find the global minimum, the mission planning hierarchy also includes a mechanism to escape from the local minimum RSSI positions.

3.2. ES-RSSO Algorithm

The ES-RSSO algorithm primarily consists of two filters: an output compensator $C_o$ and an input compensator $C_i$, as shown in Figure 5. The $s{C}_o$ filter primarily removes the white noise $n$ from the incoming RSSI link capacity signal $R$, while the $C_i/s$ filter integrates the incoming demodulated signal $\xi$ with $a_1\mathit{sin}\left({\omega }_{1}t-{\Phi }_1\right)$. The value of $\Theta$, which is calculated by summing the $a_2\mathit{sin}\left({\omega }_{2}t-{\Phi }_2\right)$ and the output signal from the $C_i/s$ filter, $\widehat{\Theta }$, is used to find the gradient of the link capacity in the RSSI map. Depending on the sign of $\alpha$, $\widehat{\Theta }$ directs the UAV either towards increasing RSSI link capacity or in the opposite direction. Using the outcome of the ES-RSSO algorithm, $\Theta$, we calculate the RSSI link capacity $r$ by continuously updating the 3D RSSI link capacity map. This process repeats until the UAV reaches the destination waypoint and lands.

In Figure 5, since $r_{t+1}=r_t\left({\Theta }_t\right)$, we can transform this into the Laplace domain as $R\left(s\right)=R\left(\Theta \left(s\right)\right)$. Thus, the calculated UAV position, $\Theta$, can be represented in the Laplace domain as follows:

$$\begin{array}{cc}\Theta \left(s\right)=& C_i\left(s\right)C_o\left(s\right)\left\{R\left(\Theta \left(s\right)\right)+N\left(s\right)\right\}{a}_1\left\{{\displaystyle \frac{{\omega }_1}{{\omega }_1^2+s^2}}{\mathit{cos}\Phi }_1+{\displaystyle \frac{{\omega }_1}{{\omega }_1^2+s^2}}{sin\Phi }_1\right\}+\hfill \\ & a_2\left\{{\displaystyle \frac{{\omega }_2}{{\omega }_2^2+s^2}}{\mathit{cos}\Phi }_2+{\displaystyle \frac{{\omega }_2}{{\omega }_2^2+s^2}}{sin\Phi }_2\right\}\hfill \end{array}$$

(2)

3.3. UAV Dynamics

A quadrotor-type UAV is used throughout the paper. Simple proportional–integral–differential (PID) controllers are applied to control the $x$, $y$, and $z$ positions, as well as the pitch, roll, and yaw angles, respectively [39]. The quadrotor UAV dynamics used in this paper are as follows:

$$\begin{array}{l}\left[\begin{array}{c}\ddot{x}\\ \ddot{y}\\ \ddot{z}\end{array}\right]=\left[\begin{array}{c}\left(\mathit{cos}\varphi \mathit{sin}\theta \mathit{cos}\psi +\mathit{sin}\varphi \mathit{sin}\psi \right){\displaystyle \frac{1}{m}}\\ \left(\mathit{cos}\varphi \mathit{sin}\theta \mathit{sin}\psi -\mathit{sin}\varphi \mathit{cos}\psi \right){\displaystyle \frac{1}{m}}\\ \left(\mathit{cos}\varphi \mathit{cos}\theta \right){\displaystyle \frac{1}{m}}\end{array}\right]U_1+\left[\begin{array}{c}0\\ 0\\ -g\end{array}\right],\\ \left[\begin{array}{c}\ddot{\varphi }\\ \ddot{\theta }\\ \ddot{\psi }\end{array}\right]=\left[\begin{array}{c}\dot{\theta }\dot{\psi }\left({\displaystyle \frac{I_y-I_z}{I_x}}\right)-{\displaystyle \frac{J_r}{I_x}}\dot{\theta }\left(w_1+w_3-w_2-w_4\right)+{\displaystyle \frac{l}{I_x}}{U}_2\\ \dot{\varphi }\dot{\psi }\left({\displaystyle \frac{I_z-I_x}{I_y}}\right)-{\displaystyle \frac{J_r}{I_y}}\dot{\varphi }\left(w_1+w_3-w_2-w_4\right)+{\displaystyle \frac{l}{I_y}}{U}_3\\ \dot{\varphi }\dot{\theta }\left({\displaystyle \frac{I_x-I_y}{I_z}}\right)+{\displaystyle \frac{1}{I_z}}{U}_4\end{array}\right],\end{array}$$

(3)

where $x$, $y$, and $z$ represent the UAV position ($\mathrm{m}$), $\varphi$, $\theta$, and $\psi$ are the roll, pitch, and yaw angles ($\mathrm{r}\mathrm{a}\mathrm{d}$), $m$ is the UAV mass ($\mathrm{k}\mathrm{g}$), $g$ is the gravitational acceleration ($\mathrm{m}/{\mathrm{s}}^2$), $I_x$, $I_y$, and $I_z$ are the body inertias (${\mathrm{k}\mathrm{g}·\mathrm{m}}^2$), $J_r$ is the rotor inertia (${\mathrm{k}\mathrm{g}·\mathrm{m}}^2$), and $l$ is the arm length ($\mathrm{m}$). The system’s inputs ($U_1$, $U_2$, $U_3$, $U_4$, and $\Omega$) can be rewritten as follows:

$$\begin{array}{l}{U}_1=b\left({\Omega }_1^2+{\Omega }_2^2+{\Omega }_3^2+{\Omega }_4^2\right),\\ U_2=b\left({\Omega }_4^2-{\Omega }_2^2\right),\\ U_3=b\left({\Omega }_3^2-{\Omega }_1^2\right),\\ U_4=d\left({\Omega }_2^2+{\Omega }_4^2-{\Omega }_1^2-{\Omega }_3^2\right),\\ \Omega ={\Omega }_2+{\Omega }_4-{\Omega }_1-{\Omega }_3,\end{array}$$

(4)

where $U$ is the input vector ($\mathrm{k}\mathrm{g}·\mathrm{m}/{\mathrm{s}}^2$), $\Omega$ is the rotor speed ($\mathrm{r}\mathrm{a}\mathrm{d}/\mathrm{s}$), $b$ is the thrust coefficient (dimensionless), and $d$ is the drag coefficient (dimensionless).

4. Simulation Result

All simulations are performed in the MATLAB/Simulink environment, as shown in Figure 6a, using a quadrotor vehicle with the following initial system properties: ${\Theta }_{init}=[0, 0, 0] \mathrm{m}$, ${\left[\varphi , \theta , \psi \right]}_{init}=[0, 0, 0] \mathrm{r}\mathrm{a}\mathrm{d}$, ${\left[\dot{\varphi }, \dot{\theta }, \dot{\psi }\right]}_{init}=[0, 0, 0] \mathrm{r}\mathrm{a}\mathrm{d}/\mathrm{s}$, $v_{init}=[0, 0, 0] \mathrm{m}/\mathrm{s}$, $a_{init}=\left[0, 0, 0\right] \mathrm{m}/{\mathrm{s}}^2$, ${‖v_{min}‖}_2=0 \mathrm{m}/\mathrm{s}$, ${‖v_{max}‖}_2=10 \mathrm{m}/\mathrm{s}$, ${Altitude}_{min}=1 \mathrm{m}$, and ${Altitude}_{normal}=10 \mathrm{m}$. The time step used is 0.01 $\mathrm{s}$. The simulations were run on a desktop with an Intel^® Core™ i5-4590 CPU (3.30 $\mathrm{G}\mathrm{H}\mathrm{z}$), a 64-bit operating system, and 8.00 $\mathrm{G}\mathrm{B}$ of RAM.

In Figure 6b, the RSSI profile over the soccer field (image sourced from Google Earth) was obtained by flying a quadrotor UAV at a height of 10 $\mathrm{m}$ under the wind speed of approximately 2 $\mathrm{m}/\mathrm{s}$. For the flight experiment, we used a PX4 Pro-powered F450 quadrotor [40] equipped with a three-layered IMU-based Pixhawk 2.1 FC, an RFD900x 915 $\mathrm{M}\mathrm{H}\mathrm{z}$ telemetry modem, QGroundControl [41], a Pixhawk 2.1 Here+ RTK [42], and a 4s 3300 $\mathrm{m}\mathrm{A}\mathrm{h}$ battery pack. These components were used in the outdoor autopilot experiment, which followed a pre-established energy-efficient path. Notably, the Pixhawk 2.1 Here+ RTK was utilized to achieve centimeter-level accuracy in autopilot path following. We saved the telemetry radio data collected during the quadrotor’s flight and extracted the RSSI data from it. Additionally, we converted the GPS coordinates to Earth-Centered Earth-Fixed (ECEF) coordinates. The converted GPS data were used for the x and y positions, while the extracted RSSI data were used for the z-axis.

For the simulation experiments, the quadrotor UAV is set to fly from ${\Theta }_{init}=[0, 0, 0] \mathrm{m}$ to ${\Theta }_{end}=[45, 32.5, 10] \mathrm{m}$ in the ECEF coordinates, as shown in Figure 7. In Figure 6b, the UAV is represented by circles in four different colors, with each color corresponding to a rotor of the quadrotor. The GCS, depicted as a red circle in Figure 7a, is represented as a yellow rectangle in Figure 6b at the center of the soccer field.

According to Figure 8a, the RSSI value tends to increase over time, indicating that the quadrotor UAV slightly modifies its trajectory as it flies towards the final waypoint. Since the ES-RSSO algorithm is designed to find the local minimum, the UAV occasionally falls into local RSSI minimum positions, as reflected by the fluctuating RSSI values shown in Figure 8b.

Lastly,

Table 2. Comparison simulation result of with and without ES-RSSO algorithm application.

	Mean	Variance	Integration
w/ES-RSSO ($\mathbf{d}\mathbf{B}\mathbf{m}$)	182.1763	12.3866	850.1319
w/o ES-RSSO ($\mathbf{d}\mathbf{B}\mathbf{m}$)	186.5946	32.6610	882.9339
Improvement ($\%$)	2.37	62.08	3.72

summarizes the simulation results, showing a 2.37% decrease in the mean, a 62.08% improvement in variance, and a 3.72% decrease in the integration strength of the link capacity when the ES-RSSO is applied. The results show that the RADIO.rssi remains consistently above the critical threshold, guaranteeing reliable communication links and facilitating energy-efficient path planning when the ES-RSSO is implemented.

The integration strength of the link capacity represents the area under each graph in Figure 8a. Additionally, Figure 8b illustrates the RSSI variance during the UAV’s flight.

5. Conclusions

In this paper, we demonstrated a method for finding the minimum RSSI trajectory using the ES-RSSO algorithm. Although the ES-RSSO algorithm occasionally falls into local minimum RSSI values, it generally performs well. It guides the UAV to follow an increasing RSSI value during mission planning. According to the simulation results, the ES-RSSO algorithm calculates gradually varying position trajectories, ensuring that the UAV’s attitude remains stable while maintaining the minimum required RSSI level.

In contrast to most previously published path optimization literature, which primarily focuses on optimal path planning such as the shortest or fastest route, the minimum RSSI trajectory proposed by the ES-RSSO algorithm helps to reduce unnecessary energy consumption in modulating and demodulating MAVLink messages by finding the optimal positions of the UAV using radio communication signals. This ensures energy-efficient path planning while guaranteeing the minimum RSSI capacity, which eventually results in the extended mission flight time.

In the future, to address the semi-global convergence issue of the proposed ES-RSSO algorithm, we plan to explore the integration of techniques such as SCA, RRT, QiER-based DRL, and IRL-based DRL with ES-RSSO. The increased computation time resulting from a more complex algorithm set will be mitigated by using a high-performance, graphics card-based mission computer alongside the FC. Alternatively, a weighted centroid localization algorithm can be applied to assign different weights to each computation algorithm depending on the mission scenarios to effectively address the trade-offs of each involved algorithm.

We will conduct additional flight experiments using an F450 quadrotor equipped with a Pixhawk FC, Here+ RTK GPS, and other components [43] to construct a 3D RSSI map while simultaneously operating the ES-RSSO algorithm to find the optimal path that results in the minimum RSSI trajectory for energy-efficient flight. We plan to introduce abrupt variations in RSSI in real-time, simulating environmental disturbances. With real-time 3D RSSI mapping and an improved ES-RSSO algorithm, we aim to achieve more energy-efficient, minimum RSSI capacity-guaranteed path planning.