# Towards 6G IoT: Tracing Mobile Sensor Nodes with Deep Learning Clustering in UAV Networks

^{1}

^{2}

^{3}

^{4}

^{5}

^{6}

^{*}

## Abstract

**:**

## 1. Introduction

- We introduced a graph representation of the UAV network that enabled the utilisation of graph convolutional network (GCN) architectures;
- The clusterisation was performed in the network at regular intervals based on the RSSI and the current locations of the UAVs, with a dynamic number of clusters at each phase;
- We utilised the optimisation of an RSSI deep learning loss function to determine the final clusters;
- The introduced method allowed the UAVs to trace the target sensor node without explicit knowledge of its location or the use of distance estimates.

## 2. Related Work

## 3. Methodology

#### 3.1. Tracing Algorithm

#### 3.2. Deep Learning Clustering for the Determination of UAV Groups

#### 3.2.1. Graph Representation of the Network

**I**is the identity matrix and

**D**and

**A**are the degree and adjacency matrices, respectively. In addition, a new matrix was created, holding information about each UAV pertaining to the location coordinates and the RSSI from the target. Using the above matrices, the final GCN model was constructed and trained to provide the optimal clusters.

#### 3.2.2. Deep Learning Loss Formulation

**Y**is the output matrix from the neural network and

**r**is a vector containing the RSSI values. By optimising this loss function, the deep learning model determined the partitions and where each node belonged. To ensure that the number of UAVs was balanced in each cluster, an additional loss function was used:

#### 3.2.3. Determining Optimal Number of Clusters

## 4. Simulation Setup

- In the first scenario, the speed of the UAVs was set to 40 km/h, and the target sensor was placed in a random position, 2 km away from the UAV deployment location;
- In the second scenario, the target was positioned 3 km away from the deployment location, and the UAVs’ velocity was set to 50 km/h.

## 5. Evaluation

#### 5.1. Results

#### 5.2. Ablation Study

## 6. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## References

- Sekander, S.; Tabassum, H.; Hossain, E. Multi-Tier Drone Architecture for 5G/B5G Cellular Networks: Challenges, Trends, and Prospects. IEEE Commun. Mag.
**2018**, 56, 96–103. [Google Scholar] [CrossRef] [Green Version] - Mozaffari, M.; Saad, W.; Bennis, M.; Debbah, M. Drone Small Cells in the Clouds: Design, Deployment and Performance Analysis. In Proceedings of the 2015 IEEE Global Communications Conference (GLOBECOM), San Diego, CA, USA, 6–10 December 2015; pp. 1–6. [Google Scholar] [CrossRef] [Green Version]
- Mozaffari, M.; Saad, W.; Bennis, M.; Nam, Y.H.; Debbah, M. A Tutorial on UAVs for Wireless Networks: Applications, Challenges, and Open Problems. IEEE Commun. Surv. Tutor.
**2019**, 21, 2334–2360. [Google Scholar] [CrossRef] [Green Version] - Lagkas, T.; Argyriou, V.; Bibi, S.; Sarigiannidis, P. UAV IoT Framework Views and Challenges: Towards Protecting Drones as “Things”. Sensors
**2018**, 18, 4015. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Zhan, C.; Zeng, Y.; Zhang, R. Energy-Efficient Data Collection in UAV Enabled Wireless Sensor Network. IEEE Wirel. Commun. Lett.
**2018**, 7, 328–331. [Google Scholar] [CrossRef] [Green Version] - Amponis, G.; Lagkas, T.; Sarigiannidis, P.; Vitsas, V.; Fouliras, P. Inter-UAV Routing Scheme Testbeds. Drones
**2021**, 5, 2. [Google Scholar] [CrossRef] - Mozaffari, M.; Saad, W.; Bennis, M.; Debbah, M. Mobile Unmanned Aerial Vehicles (UAVs) for Energy-Efficient Internet of Things Communications. IEEE Trans. Wirel. Commun.
**2017**, 16, 7574–7589. [Google Scholar] [CrossRef] - Zong, B.; Fan, C.; Wang, X.; Duan, X.; Wang, B.; Wang, J. 6G Technologies: Key Drivers, Core Requirements, System Architectures, and Enabling Technologies. IEEE Veh. Technol. Mag.
**2019**, 14, 18–27. [Google Scholar] [CrossRef] - Saad, W.; Bennis, M.; Chen, M. A Vision of 6G Wireless Systems: Applications, Trends, Technologies, and Open Research Problems. IEEE Netw.
**2020**, 34, 134–142. [Google Scholar] [CrossRef] [Green Version] - Xiao, Z.; Zeng, Y. An Overview on Integrated Localization and Communication Towards 6G. arXiv
**2020**, arXiv:2006.01535. [Google Scholar] - Aazhang, B.; Ahokangas, P.; Alves, H.; Alouini, M.S.; Beek, J.; Benn, H.; Bennis, M.; Belfiore, J.; Strinati, E.; Chen, F.; et al. Key Drivers and Research Challenges for 6G Ubiquitous Wireless Intelligence; White Paper; 6G Flagship: Oulu, Finland, 2019. [Google Scholar]
- Alsheikh, M.A.; Lin, S.; Niyato, D.; Tan, H. Machine Learning in Wireless Sensor Networks: Algorithms, Strategies, and Applications. IEEE Commun. Surv. Tutor.
**2014**, 16, 1996–2018. [Google Scholar] [CrossRef] [Green Version] - Shalev-Shwartz, S.; Ben-David, S. Understanding Machine Learning: From Theory to Algorithms; Cambridge University Press: Cambridge, MA, USA, 2014. [Google Scholar]
- Langley, P.; Simon, H.A. Applications of Machine Learning and Rule Induction. Commun. ACM
**1995**, 38, 54–64. [Google Scholar] [CrossRef] - Nikitas, A.; Michalakopoulou, K.; Njoya, E.T.; Karampatzakis, D. Artificial Intelligence, Transport and the Smart City: Definitions and Dimensions of a New Mobility Era. Sustainability
**2020**, 12, 2789. [Google Scholar] [CrossRef] [Green Version] - Krishnamachari, L.; Estrin, D.; Wicker, S. The impact of data aggregation in wireless sensor networks. In Proceedings of the 22nd International Conference on Distributed Computing Systems Workshops, Vienna, Austria, 2–5 July 2002; pp. 575–578. [Google Scholar] [CrossRef]
- Spyridis, Y.; Lagkas, T.; Sarigiannidis, P.; Zhang, J. Modelling and simulation of a new cooperative algorithm for UAV swarm coordination in mobile RF target tracking. Simul. Model. Pract. Theory
**2021**, 107, 102232. [Google Scholar] [CrossRef] - Gumaida, B.F.; Luo, J. Novel localization algorithm for wireless sensor network based on intelligent water drops. Wirel. Netw.
**2019**, 25, 597–609. [Google Scholar] [CrossRef] - Mehdi Dehghan, S.M.; Moradi, H.; Asghar Shahidian, S.A. Optimal path planning for DRSSI based localization of an RF source by multiple UAVs. In Proceedings of the 2014 Second RSI/ISM International Conference on Robotics and Mechatronics (ICRoM), Tehran, Iran, 15–17 October 2014; pp. 558–563. [Google Scholar] [CrossRef]
- Sarunic, P.; Evans, R. Hierarchical model predictive control of UAVs performing multitarget-multisensor tracking. IEEE Trans. Aerosp. Electron. Syst.
**2014**, 50, 2253–2268. [Google Scholar] [CrossRef] - Koohifar, F.; Kumbhar, A.; Guvenc, I. Receding Horizon Multi-UAV Cooperative Tracking of Moving RF Source. IEEE Commun. Lett.
**2017**, 21, 1433–1436. [Google Scholar] [CrossRef] - Mavrommati, A.; Tzorakoleftherakis, E.; Abraham, I.; Murphey, T. Real-Time Area Coverage and Target Localization Using Receding-Horizon Ergodic Exploration. IEEE Trans. Robot.
**2017**. [Google Scholar] [CrossRef] [Green Version] - Koohifar, F.; Guvenc, I.; Sichitiu, M.L. Autonomous Tracking of Intermittent RF Source Using a UAV Swarm. IEEE Access
**2018**, 6, 15884–15897. [Google Scholar] [CrossRef] - Pack, D.J.; DeLima, P.; Toussaint, G.J.; York, G. Cooperative Control of UAVs for Localization of Intermittently Emitting Mobile Targets. IEEE Trans. Syst. Man Cybern. Part B (Cybernetics)
**2009**, 39, 959–970. [Google Scholar] [CrossRef] [PubMed] - Faruk Gencel, M.; Madhowl, U.; Pedro Hespanhal, J. RF Source Seeking Using Frequency Measurements. In Proceedings of the 2018 IEEE 19th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), Kalamata, Greece, 25–28 June 2018; pp. 1–5. [Google Scholar] [CrossRef]
- Acuna, V.; Kumbhar, A.; Vattapparamban, E.; Rajabli, F.; Guvenc, I. Localization of WiFi Devices Using Probe Requests Captured at Unmanned Aerial Vehicles. In Proceedings of the 2017 IEEE Wireless Communications and Networking Conference (WCNC), San Francisco, CA, USA, 19–22 March 2017; pp. 1–6. [Google Scholar] [CrossRef]
- Kyösti, P.; Meinilä, J.; Hentilä, L.; Zhao, X.; Jämsä, T.; Schneider, C.; Narandzić, M.; Milojević, M.; Hong, A.; Ylitalo, J.; et al. WINNER II Channel Models Part I Channel Models; 2008. IST-4-027756 WINNER II D1.1.2 v1.2 WINNER II Channel Models. Information Society Technologies 11. Available online: https://www.google.com.hk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwj9ltrMpoXxAhVOBKYKHUu8AfIQFjACegQIAhAD&url=http%3A%2F%2Fwww.ero.dk%2F93F2FC5C-0C4B-4E44-8931-00A5B05A331B%3Fframes%3Dno%26&usg=AOvVaw0XgSCl_4J6iDdFg3TdTDxV (accessed on 6 June 2021).
- Nazi, A.; Hang, W.; Goldie, A.; Ravi, S.; Mirhoseini, A. GAP: Generalizable Approximate Graph Partitioning Framework. arXiv
**2019**, arXiv:1903.00614. [Google Scholar]

**Figure 2.**Structure of the deep learning network. The input layer’s dimensions are n × 3, where n is the number of UAVs in the UAV network. The output layer has k × 1 dimensions, with k being the number of clusters.

**Figure 3.**Example of a plot demonstrating the elbow method. The distortion score is computed as the sum of squared distances from each point to its assigned centre. In this case, the number of selected clusters should be 4.

**Figure 4.**Simulator instances at different timestamps of the simulation. The red square represents the target sensor, and the dots indicate the UAVs. The left subfigure depicts the swarm 90 s after initiating the algorithm. The middle subfigure corresponds to 120 s, while the last subfigure demonstrates the tracing process after the threshold for terminating the DL procedure has been reached.

**Figure 5.**Time required for the first UAV to reach the target sensor when placed at a distance of 2 km, versus the standard deviation $\sigma $ of the additive noise. Comparison of the proposed method at clustering termination thresholds of 20, 15 and 10 dB and the previous approach.

**Figure 6.**Time required for the first UAV to reach the target sensor when placed at a distance of 3 km, versus the standard deviation $\sigma $ of the additive noise. Comparison of the proposed method at clustering termination thresholds of 20, 15 and 10 dB and the previous approach.

**Figure 7.**Average distance covered per UAV when the target sensor is placed at a distance of 2 km, versus the standard deviation $\sigma $ of the additive noise. Comparison of the proposed method at clustering termination thresholds of 20, 15 and 10 dB and the previous approach.

**Figure 8.**Total distance covered by all UAVs when the target sensor is placed at a distance of 2 km, versus the standard deviation $\sigma $ of the additive noise. Comparison of the proposed method at clustering termination thresholds of 20, 15 and 10 dB and the previous approach.

**Figure 9.**Average distance covered per UAV when the target sensor is placed at a distance of 3 km, versus the standard deviation $\sigma $ of the additive noise. Comparison of the proposed method at clustering termination thresholds of 20, 15 and 10 dB and the previous approach.

**Figure 10.**Total distance covered by all UAVs when the target sensor is placed at a distance of 3 km, versus the standard deviation $\sigma $ of the additive noise. Comparison of the proposed method at clustering termination thresholds of 20, 15 and 10 dB and the previous approach.

**Figure 11.**Time required for the first UAV to reach the target sensor when placed at a distance of 2 km, versus the standard deviation $\sigma $ of the additive noise. Results at clustering termination thresholds of 20 dB, with and without using the elbow method to determine the number of clusters.

**Table 1.**Parameters of the simulation for each examined scenario. The two scenarios are differentiated in terms of the UAVs’ velocity and the target’s initial distance.

Parameter | Scenario 1 | Scenario 2 |
---|---|---|

Number of UAVs | 45 | 45 |

Cluster update interval | 30 s | 30 s |

UAV altitude | 100 m | 100 m |

UAV velocity | 40 km/h | 50 km/h |

Target velocity | 5 km/h | 5 km/h |

Target distance | 2 km | 3 km |

Target TX power | 10 mW | 10 mW |

Target TX antenna gain | 2 dBi | 2 dBi |

UAV RX antenna gain | 2 dBi | 2 dBi |

Signal frequency | 2400 MHz | 2400 MHz |

Total duration | 1000 s | 1000 s |

**Table 2.**Average results for all values of the standard deviation $\sigma $ of the additive noise for the DL method and the previous approach, at different thresholds. The target is placed at a 2 km distance. Time is measured in seconds and distance in meters.

Threshold | Proposed | Spyridis et al. [17] | ||||
---|---|---|---|---|---|---|

Time | Average Distance | Total Distance | Time | Average Distance | Total Distance | |

10 | 260 | 8661 | 249,032 | 332 | 9201 | 414,214 |

15 | 268 | 8351 | 126,716 | 295 | 9270 | 417,322 |

20 | 251 | 8222 | 99,191 | 268 | 9345 | 420,705 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Spyridis, Y.; Lagkas, T.; Sarigiannidis, P.; Argyriou, V.; Sarigiannidis, A.; Eleftherakis, G.; Zhang, J.
Towards 6G IoT: Tracing Mobile Sensor Nodes with Deep Learning Clustering in UAV Networks. *Sensors* **2021**, *21*, 3936.
https://doi.org/10.3390/s21113936

**AMA Style**

Spyridis Y, Lagkas T, Sarigiannidis P, Argyriou V, Sarigiannidis A, Eleftherakis G, Zhang J.
Towards 6G IoT: Tracing Mobile Sensor Nodes with Deep Learning Clustering in UAV Networks. *Sensors*. 2021; 21(11):3936.
https://doi.org/10.3390/s21113936

**Chicago/Turabian Style**

Spyridis, Yannis, Thomas Lagkas, Panagiotis Sarigiannidis, Vasileios Argyriou, Antonios Sarigiannidis, George Eleftherakis, and Jie Zhang.
2021. "Towards 6G IoT: Tracing Mobile Sensor Nodes with Deep Learning Clustering in UAV Networks" *Sensors* 21, no. 11: 3936.
https://doi.org/10.3390/s21113936