Merge overleaf-2024-11-06-1042 into main

article.hid2-ub-cicd.ppam24.tex

@@ -282,7 +282,7 @@ \subsubsection{Computational Tools for Meshes Generation}
 All the features described above are not yet operational in our framework. We can reconstruct a geometric model of an urban area whose location is arbitrary so that the user can select an area anywhere in the world. The final mesh comprises all buildings LOD-0 or LOD-1 and the terrain with elevation. The integration of all the urban components is currently being investigated. This part requires fairly costly algorithms (mesh intersection), for which it will be necessary to form groups intelligently to reduce the computational cost.
 
 \paragraph{Current Mesh Generation Strategy}
-The current strategy for mesh generation employs multi-threading (MT) to handle various stages of the mesh construction process. Initially, the tiles used are determined based on the specified location, a task performed sequentially. Following this, GIS data, including buildings and elevation information, is downloaded in parallel. Next, polygons are repaired~\cite{loriot_polygon_2024} to ensure they are suitable for mesh generation, with this process executed in parallel using MT. The terrain mesh is then generated in parallel, employing CGAL's algorithms to ensure precision and efficiency. Union operations at tile junctions are performed to ensure continuity, a step that is currently sequential. Finally, building meshes are generated using a parallel MT approach, leveraging CGAL for its advanced mesh generation capabilities.
+The current strategy for mesh generation employs multi-threading (MT) to handle various stages of the mesh construction process. Initially, the used tiles are determined based on the specified location, a task performed sequentially. Following this, GIS data, including buildings and elevation information, is downloaded in parallel. Next, polygons are repaired~\cite{loriot_polygon_2024} to ensure they are suitable for mesh generation, with this process executed in parallel using MT. The terrain mesh is then generated in parallel, employing CGAL's algorithms to ensure precision and efficiency. Union operations at tile junctions are performed to ensure continuity, a step that is currently sequential. Finally, building meshes are generated using a parallel MT approach, leveraging CGAL for its advanced mesh generation capabilities.
 
 \paragraph{Advancing Towards Full Parallelism}
 The following steps in enhancing the mesh generation process involve moving towards a fully parallel approach using both Multi-threading (MT) and Message Passing Interface (MPI). The goal is to scale the process to handle entire cities or larger urban areas. This scale-up involves utilizing MPI to manage distributed computing resources effectively. To achieve the latter objective, partitions of tiles with overlapping regions are created to ensure complete and accurate building descriptions without requiring extensive MPI communications. Furthermore, each process uses an MT strategy to generate meshes independently, with overlapping zones allowing for the creation of complete building structures.