Abstract
The cooperative localization of Unmanned Aerial Vehicles (UAVs) has emerged as a pivotal application in Internet of Things (IoT) tasks. However, the frequent exchange of localization data among UAVs leads to significant energy consumption and escalates the computational complexity involved in multi-UAV cooperative localization tasks. To address these challenges, this paper proposes a cooperative localization algorithm that integrates a biogeography optimization-based cluster networking and adaptive sampling-improved Nystrom super-multidimensional scaling (BOCN-ASNSMS). The proposed method leverages biogeography optimization (BO), prioritizing nodes with higher residual energy and density to serve as cluster heads, thereby optimizing energy usage. Subsequently, an improved adaptive sampling Nystrom super-multidimensional scaling algorithm is employed to dynamically select the kernel matrix row vectors. This selection process not only reduces data processing requirements but also enhances the accuracy of the similarity matrix approximation, thus diminishing computational complexity and achieving precise relative positioning of UAVs. Furthermore, Procrustes analysis and least squares methods are utilized to fuse coordinates across UAV clusters, aligning them into a unified coordinate system and converting them into absolute coordinates, which facilitates high-precision global localization. Theoretical analysis and simulation results underscore that the proposed algorithm substantially reduces computational complexity and energy consumption while enhancing localization accuracy, compared to conventional multi-UAV cooperative localization approaches.