Safdar, Zanab (2025) AI-Enabled Internet of Trees (IoTr) Simulation Framework for Real-Time Monitoring and detection of tree diseases in the United Kingdom. Doctoral thesis, University of West London.

PDF (PDF/A) AI-Enabled Internet of Trees (IoTr) Simulation Framework_Zanab Safdar _Thesis (Sept 2025)_accessible.pdf - Submitted Version Restricted to Repository staff only until September 2026. Available under License Creative Commons Attribution. Download (11MB) | Request a copy

Abstract

The Internet of Things (IoT) has introduced transformative capabilities for ecological monitoring through the acquisition of real-time sensor-based data. In this context, the Internet of Trees (IoTr) has emerged as a promising domain for monitoring trees in forest ecosystems. However, the existing literature lacks species-specific, energy-efficient, and scalable frameworks capable of supporting proactive tree health surveillance and disease detection. Current research focuses on using IoTr for tree monitoring, but primarily covers basic soil moisture, with limited integration of disease-relevant parameters and species classification. This research addresses the existing research gap by proposing IoT-Based Tree Routing for Energy-Efficient Systems (I-TREES), a novel context-aware framework for real-time monitoring of tree health, with a specific focus on ash dieback and oak decline in the United Kingdom (UK). This research first evaluates three long-range communication protocols: Narrowband IoT (NB-IoT), Long Range Wide Area Network (LoRaWAN), and Sigfox, using MATLAB simulations to compare energy efficiency, Packet Delivery Ratio (PDR), Packet Loss Ratio (PLR), delay, latency, and throughput across two node densities (100 and 300 nodes). Based on the comparative evaluation, NB-IoT was selected as the optimal protocol, and the I-TREES framework is designed and simulated in Network Simulator 3 (NS-3) using a large-scale network of 700 IoT nodes to assess its performance in forest health monitoring. The comparative benchmark against six established routing protocols Low-Energy Adaptive Clustering Hierarchy (LEACH), Power-Efficient Gathering in Sensor Information Systems (PEGASIS), Hierarchical Clustering Algorithm (HCA), Heuristic Clustering and Scheduling Algorithm (H-CSA), Event-Driven Spatio-Temporal Optimisation (ESTO), and K-Means. The results showed that I-TREES consistently outperformed in terms of scalability, throughput, latency, and energy efficiency. In the final phase, a Proof of-Concept (PoC) approach is adopted to assess the feasibility of disease classification under limited real-world data conditions. A multimodal PoC dataset was constructed by integrating IoT sensor readings with custom labelled image data of ash and oak trees. Machine Learning (ML) models, including Random Forest (RF), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and hybrid (CNN-RF, CNN-LSTM) architectures, are trained for disease detection. The RF-based multimodal model, in particular, offered strong interpretability and demonstrated consistently high performance in accuracy, precision, recall, and F1 score, supporting its suitability for real world deployment. The findings verified the effectiveness of the framework as an energy efficient and scalable approach for the detection of tree diseases. The proposed I-TREES framework presents a reliable and intelligent forest health surveillance framework for tree disease detection, suitable for real-world forest applications to manage diseases sustainably and support policy-making decisions for ecological resilience in the UK and similar environments.

Item Type:	Thesis (Doctoral)
Identifier:	10.36828/thesis/14390
Keywords:	Internet of Things (IoT); Internet of Trees (IoTr); Wireless Sensor Networks (WSN); Low Power Wide Area Networks (LPWAN); Narrowband IoT (NB-IoT); Smart Environmental Monitoring; Multimodal Data Fusion; Artificial Intelligence (AI); Machine Learning (ML); Deep Learning (DL); Random Forest (RF); Convolutional Neural Networks (CNN); Long Short-Term Memory (LSTM)
Subjects:	Construction and engineering > Digital signal processing Natural sciences
Date Deposited:	03 Dec 2025
URI:	https://repository.uwl.ac.uk/id/eprint/14390

Actions (admin access)