HAPS research shows 94% reduction in GPS error using only 4 platforms

High-altitude platform stations (HAPS) – originally built for communications – can double as signals of opportunity to augment GPS in urban canyons where satellite signals suffer from blockage and non-line-of-sight conditions. However, naive HAPS placement yields limited gains. A new paper on arXiv by Zheng, Atia, and Yanikomeroglu introduces a metaheuristic framework that jointly optimizes the number and placement of HAPS under real-world constraints. The framework integrates high-fidelity 3D city models, ray-tracing, and multi-objective optimization to handle the discrete, non-convex design space. The authors developed three different metaheuristic solvers (e.g., genetic algorithms, particle swarm variants) that all converged rapidly and outperformed a greedy baseline, especially in low-to-moderate HAPS regimes.

For representative dense urban scenarios, the optimal configuration required only four HAPS to meet an 18-meter average 3D positioning error bound (PEB). Configurations with two to five HAPS achieved over 50% reduction in mean and root-mean-square PEB compared to satellite-only solutions. More strikingly, the standard deviation of error dropped by up to 94% and the coefficient of variation by up to 87%, meaning the system is both more accurate and far more consistent. Beyond six HAPS, geometric redundancy caused diminishing returns, underscoring that smarter placement matters more than just adding platforms. The framework also proved robust across different city morphologies and propagation conditions, making it a scalable solution for real-world HAPS-aided localization. This research could accelerate deployment of HAPS as a reliable backup or augmentation layer for critical navigation in urban environments.