Comparative Study for Coordinating Multiple Unmanned HAPS for Communications Area Coverage

Abstract

This work compares the application of Reinforcement Learning (RL) and Swarm Intelligence (SI) based methods for resolving the problem of coordinating multiple High Altitude Platform Stations (HAPS) for communications area coverage. Swarm coordination techniques are essential for developing autonomous capabilities for multiple HAPS/UAS control and management. This paper examines the performance of artificial intelligence (AI) capabilities of RL and SI for autonomous swarm coordination. In this work, it was observed that the RL approach showed superior overall peak user coverage with unpredictable coverage dips; while the SI based approach demonstrated lower coverage peaks but better coverage stability and faster convergence rates.

A Survey on Machine-Learning Techniques for UAV-Based Communications

“…Various works leveraged swarm-based optimization tools in UAV networks. In Reference [70], a network with multiple high amplitude platform stations (HAPS) is considered for communication coverage purposes. The coordination of this network has been performed with the aid of RL and swarm intelligence.…”

Implementing Solar-Powered HAPS for Rural broadband Connectivity: Concepts, Challenges & Mitigation

“…The use of manual methods (direct human input) to coordinate such a network will entail considerable operational complexity and cost. Consequently, the authors of this paper have been examining the use of algorithms like Reinforcement Learning (RL) and Swarm Intelligence (SI) to coordinate multiple fixed-wing solar-HAPS for communications coverage [18]. The challenges of coordinating multiple fixed-wing solar-HAPS differ significantly from applications using balloons e.g.…”

Situation Awareness and Routing Challenges in Unmanned HAPS/UAV based Communications Networks

“…Regardless, these types of UAV based networks are plagued by challenges of high mobility and frequent topology changes. These challenges ultimately make designing an appropriate routing protocol a very difficult task [5], [6].…”

“…A different design paradigm or approach may be needed to address routing challenges peculiar to UAV networks; merely modifying traditional or MANET specific protocols may not solve the problem. For instance, in an earlier paper [6], the authors proposed integrating routing and UAV autonomy algorithms to mitigate link availability challenges for improved routing performance. In order to address routing challenges, innovative and proactive approaches are needed, even if it leads to adjusting design architectures.…”

“…The application of SA to routing may be a departure from current approaches which generally limits its scope to addressing layer 3 elements (a purely organic approach). The approach considered in this paper assumes a more external approach by including the UAV autonomy algorithm (SA resides here) in the UAV decision making framework as proposed by Anicho et al [6]. This design approach ensures that at the conceptual level UAV autonomy and the routing algorithm is addressed simultaneously.…”