Self-Aware Intelligent Model (SAIMOD) – A fault-tolerant UAV-based communication system for Disaster Recovery

Erim, O (2025) Self-Aware Intelligent Model (SAIMOD) – A fault-tolerant UAV-based communication system for Disaster Recovery. Doctoral thesis, Liverpool John Moores University.

Preview

Text 2025 ERIM PhD.pdf - Published Version Available under License Creative Commons Attribution Non-commercial. Download (4MB) | Preview

Abstract

When disasters such as landslides, floods, earthquakes, forest fires or avalanches occur, preserving human lives is the most critical issue that needs to be solved. As such, Search and Rescue (SAR) operations must begin quickly and efficiently. A key aspect of SAR missions is creating or re-establishing the destroyed or damaged communication links/infrastructure in the disaster recovery area. There is usually a demand for communication links between the entities involved, such as people to be rescued and rescue team members. One possible solution that has gained traction over the years is ad hoc networks. These networks allow the concerned entities, namely people to be rescued, rescue team members, etc., to form decentralised communication links quickly by using common devices like PDAs and mobile phones. As the backbone communication infrastructure might be missing or damaged within the affected area, an external means of providing a temporary communication infrastructure must be considered.

This Research proposes a Self-Aware Intelligent Model (SAIMOD) as a framework designed to be an integrated emergency communication system that relies on Mobile Ad Hoc Networks (MANETs) formed by Unmanned Aerial Vehicles (UAVs) deployed in an example scenario within disaster recovery. The UAVs would form auxiliary MANETs, connecting the rescue teams and the people to be rescued (survivors). Each deployed UAV acts as a ‘router/relay’ to restore bi-directional connectivity between the ground nodes made up of the rescue team and the people to be rescued. It could also be used to gather and send crucial information. The MANETs formed would allow data transmission in an energy-efficient and timely manner. Intelligent networking amongst the deployed UAVs will ensure adequate coverage of the disaster area and fault-tolerant UAV formation control in real-time.

SAIMOD demonstrates self-awareness in its network resilience, allowing it to quickly adapt to unexpected changes in the UAV-based Mobile Ad hoc Network (MANET), such as a UAV failure. In the context of this framework, resilience refers to the system’s ability to maintain functionality and adapt to unexpected changes or failures within the UAV-based MANET. It utilises reinforcement learning based on a path-planning algorithm to optimise coverage and connectivity among the UAVs and ground nodes. SAIMOD is designed to adapt to real-time network changes, like a UAV failure. By employing reinforcement learning, the UAVs within SAIMOD can apply knowledge gained from one task to another. This approach ensures seamless connectivity and networking among the UAVs while preventing duplicated efforts to cover the network.

A suitable routing protocol is used in combination with a designated mobility model. Four mobility models (Random Waypoint mobility model, Reference Point Group Mobility Model (RPGMM), Manhattan and Gauss Markov mobility models) were integrated with five different routing protocols (AODV, DSR, DSDV, TORA and OLSR) in similar UAV-deployed MANETs to select the best-fit routing protocol and best-fit mobility model for use in SAIMOD via simulation. The networks formed were evaluated for a range of given metrics, and the Reference Point Group Mobility Model and the OLSR protocol were selected based on desired network performance metrics such as real-time connectivity, Throughput, MOS (Mean Opinion Score) value, etc. OLSR was further optimised for energy conservation in a two-layer modification process. This modified OLSR, MODOLSR, a novel protocol, is used with the RGPMM in SAIMOD.

This study also presents a mathematical model for using the proposed framework as an assessment model for resource management. It compares the proposed framework, the Self-Aware Intelligent Model (SAIMOD), with similar existing systems.

Keywords: UAV-based MANETs, Disaster Recovery Framework, Disaster Management, Routing, Fault-Tolerant Control, Energy conservation

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	Disaster Recovery Framework; Disaster Management; Energy conservation; UAV-based MANETs; Fault-Tolerant Control; Mobility Models and Routing protocols
Subjects:	T Technology > TA Engineering (General). Civil engineering (General)
Divisions:	Engineering
Date of acceptance:	16 April 2025
Date of first compliant Open Access:	23 June 2025
Date Deposited:	23 Jun 2025 09:56
Last Modified:	23 Jun 2025 09:56
DOI or ID number:	10.24377/LJMU.t.00026355
Supervisors:	Johnson, P and Jones, K
URI:	https://researchonline.ljmu.ac.uk/id/eprint/26355

View Item