Geometry-Optimized Fission Clustering for Joint Communication and Localization in Capacity-Constrained Networks

Clustering is a crucial technology for addressing the challenges of dense wireless networks with limited base station (BS) capacity. However, conventional clustering algorithms often neglect connectivity, leading to severe resource waste during sub-band allocation. Furthermore, they ignore the geometric distribution of nodes, frequently resulting in collinear anchors or coplanar anchor-agent configurations that cause the Equivalent Fisher Information Matrix (EFIM) to become singular, thereby degrading 3D positioning accuracy. To address these challenges, we propose the Geometry-Optimized Fission Clustering (GOFC) algorithm. GOFC constructs a multi-dimensional similarity graph incorporating a topological volumetric similarity metric to prioritize highly divergent, volumetric node groups. By employing a recursive spectral fission mechanism, GOFC guarantees connectivity and adheres to cluster capacity limits. Simulation results demonstrate that, compared with benchmark methods, the proposed GOFC algorithm attains higher network throughput while simultaneously achieving a significantly lower Squared Position Error Bound (SPEB) in highly dynamic environments.