Leveraging XAI and AWS IoT for Secure and Efficient LoRaWAN-Based Drone Fleet Operations
Supervisor: Dr. Arshad Farhad, Assistant Professor
Co-Supervisor: Dr. Khurram Ehsan, Snr. Associate Professor
Campus/School/Dept: BUIC-E8/BSEAS/CS
Supervisory Record:
- PhD Produced: Nil
- PhD Enrolled: Nil
- MS/MPhil Produced: Nil
- MS/MPhil Enrolled: 02
Topic Brief Description:
The relentless growth of unmanned aerial vehicles (UAVs), in environmental monitoring applications, necessitates the development of efficient, secure, and scalable communication networks. LoRaWAN, emerges as a promising solution for its long-range and low-power characteristics, making it well-suited for transmitting sensor data from UAVs. However, the increasing density of drone networks using LoRaWAN presents challenges in resource management and susceptibility to cyberattacks, particularly GPS spoofing. This research will utilize AI-powered solution that optimizes LoRaWAN resource utilization and enhances drone network security. The research will leverage TinyML, reinforcement learning, and federated learning to adjust network parameters for efficiency and resilience dynamically. The proposed platform will use LoRaWAN connectivity for long-range data collection, which is then aggregated in the Amazon Web Service cloud and visualized using Grafana. Additionally, the proposed design will implement AI-based anomaly detection mechanisms to identify and mitigate GPS spoofing threats, ensuring the integrity of environmental monitoring data.
Figure 1. Illustration of the AI-powered LoRaWAN drone network designed for efficient environmental monitoring. UAVs equipped with sensors, collecting data, transmitted via LoRaWAN to the cloud. AWS services, including IoT Core, Lambda, and Analytics, process and analyze the data to provide valuable insights.
Keywords: LoRaWAN, UAV, IoT, AI, Machine Learning, Reinforcement Learning, Federated Learning, Cybersecurity, GPS Spoofing, Environmental Monitoring, Cloud Computing, AWS, Edge Computing, TinyML.
Research Objectives/Deliverables:
- To design and implement a secure and efficient LoRaWAN network capable of supporting a large number of UAVs for environmental monitoring applications, integrating AWS IoT services to facilitate seamless data collection, processing, and analysis, as well as enabling remote monitoring and control of UAVs.
- To employ TinyML/deep learning/reinforcement learning techniques to dynamically adjust network parameters, such as transmission power and data rates, to maximize network efficiency and minimize interference.
- Implement AI-based anomaly detection mechanisms to identify and mitigate GPS spoofing attacks, protecting the integrity of environmental data.
Research Questions:
- How can TinyML be leveraged to optimize LoRaWAN network parameters for efficient resource utilization and improved network performance?
- What are the most effective AI-based techniques for detecting and mitigating GPS spoofing attacks in LoRaWAN-based drone networks?
- How can federated learning be applied to train AI models on edge devices to enhance privacy and security while improving the overall performance of the network?
Candidate’s Eligibility Profile:
- The applicant must have an MS/MPhil/Equivalent degree in Computer Science, Data Science, Cybersecurity or related field with CGPA >0. Besides, applicants must have a strong background in Machine Learning, IoT, Cloud Computing, and Cryptography.
- Experience with programming languages such, C++ and Python is advantageous. Candidates should thrive in an international environment and have excellent communication skills to actively contribute to team research efforts.
- Proficiency in spoken and written English is essential. We value independence and responsibility while promoting teamwork and collaboration among colleagues.