Artificial Intelligence-Based Autonomous UAV Networks: A Survey

Abstract

Recent advancements in unmanned aerial vehicles (UAVs) have proven UAVs to be an inevitable part of future networking and communications systems. While many researchers have proposed UAV-assisted solutions for improving traditional network performance by extending coverage and capacity, an in-depth study on aspects of artificial intelligence-based autonomous UAV network design has not been fully explored yet. The objective of this paper is to present a comprehensive survey of AI-based autonomous UAV networks. A careful survey was conducted of more than 100 articles on UAVs focusing on the classification of autonomous features, network resource management and planning, multiple access and routing protocols, and power control and energy efficiency for UAV networks. By reviewing and analyzing the UAV networking literature, it is found that AI-based UAVs are a technologically feasible and economically viable paradigm for cost-effectiveness in the design and deployment of such next-generation autonomous networks. Finally, this paper identifies open research problems in the emerging field of UAV networks. This study is expected to stimulate more research endeavors to build low-cost, energy-efficient, next-generation autonomous UAV networks.

A Survey on Non-Orthogonal Multiple Access for Unmanned Aerial Vehicle Networks: Machine Learning Approach

“…The works of [20] and [23] utilized AI-aided solutions for better beyond 5G generation by discovering that the ML-based paradigm enables cost-effective UAV network deployment through resource management, routing, and access protocols. Different AI systems were presented within three classification schemes: applications, algorithms, and training [25].…”

Models of the System of Collective Self-Organisation of Unmanned Aerial Vehicles Using Artificial Intelligence

“…A review and analysis of the literature on UAV networks has shown that AI-based UAVs are a technologically feasible and economically viable paradigm for the cost-effective design and deployment of such next-generation autonomous networks. This article identifies open research challenges in the emerging field of UAV networks [7]. Yograj Singh Mandloi & Yoshinobu Inada's research focuses on applying machine learning and neural networks to select actions and better understand the environment to control unmanned aerial vehicles, rather than using explicit models to achieve the same goal.…”

“…Methods for improving and implementing innovative technologies in UAVs' collective selforganisation systems have been studied by domestic and foreign scientists relatively recently [5,7,10,11]. Numerous scientific articles propose solutions to problematic issues in the field of AI application in UAV control systems, building a control system based on various systems and platforms, for example, using multiagent systems.…”

Comprehensive Investigation of Unmanned Aerial Vehicles (UAVs): An In-Depth Analysis of Avionics Systems

“…From another perspective, AI allows for automatic feature extraction while better managing onboard resources, which differentiates it from traditional cognitive algorithms [103]. Concerning UAV navigation, AI algorithms are characterized in this manuscript according to their paradigm as follows: AI algorithms that involve mathematical model formulation to find the best solution to a given problem by relying on predefined rules and objectives to guide the UAV are grouped in the first set (i.e., mathematical optimization) [104]. The other set emphasizes a paradigm that trains the models to make better decisions in UAV navigation (i.e., performance is evolved over time based on training data and gained experience) [105].…”