Differences

This shows you the differences between two versions of the page.

--- dsrt2025 [2025/07/16 08:52] – created gda
+++ dsrt2025 [2025/07/22 08:26] (current) – [Resources] gda
@@ Line 10: / Line 10: @@
 //
-We present a study on the intricate problem of the simulation of large scale complex networks. The complexity of the simulation is due to the need to represent networks with a high number of nodes and links, but even more to the need to model interactions among network nodes. We focus on discrete-event simulation, a simulation methodology that enables both sequential (i.e. monolithic) and parallel/distributed simulation (i.e. PADS) approaches. We discuss performance and scalability requirements that the simulator should have. We also discuss the need to have practical methods to easily build simulation models. In this sense, an agent-based simulation approach facilitates the model definition. To demonstrate the viability of this simulation technique, we focus on a tool we built to simulate complex networks. The tool exploits adaptive partitioning mechanisms, which are essential to reduce the communication overhead in the PADS. An experimental evaluation has been conducted using different network topologies and simulator setups. Results demonstrate the feasibility of the approach to simulate complex networks.
+Parallel And Distributed Simulation (PADS) is a well known modeling paradigm that allows efficient implementation of large simulation models as a collection of interacting entities, called Simulated Entities (SEs). SEs can then be assigned to independent execution units for parallel execution, if possible. Allocating SEs to execution units is one of the most challenging problems in PADS: ideally, highly interacting SEs should be placed on the same execution unit so that all communications are local; however, placing too many SEs on the same processor might degrade performance. Furthermore, many simulation models exhibit non-uniform computation and communication patterns among components, that might change during execution at unpredictable times. In this paper we propose a clustering heuristic that exploits communication locality, with the aim to reduce the communication cost experienced by the PADS during the execution. The heuristic adapts automatically to changing interaction patterns by migrating SEs, and can do so without any user-visible modification of the simulation model; there are, however, some parameters that can be used to tune the heuristic. We perform a large set of computational experiments to assess the effectiveness of the heuristic on a real-world scenario, with the aim of guiding the users in selecting optimal values of the heuristic parameters.
 //
 ====== Keywords ======
-   * Parallel and Distributed Simulation (PADS), Simulation, Complex Networks, Performance Evaluation
+   * Parallel and Distributed Simulation, Discrete Event Simulation, Model Partitioning, Clustering Heuristics.
 ====== Status ======
-   * Gabriele D'Angelo, Stefano Ferretti, Adaptive parallel and distributed simulation of complex networks, Journal of Parallel and Distributed Computing, Volume 163, 2022, Pages 30-44, ISSN 0743-7315, https://doi.org/10.1016/j.jpdc.2022.01.022.
+   * Submitted for review to the 2025 29th International Symposium on Distributed Simulation and Real Time Applications (DS-RT 2025).
 ====== Resources ======
-  * The LUNES source code is part of the [[pads:download|ARTÌS software distribution]]
+  * Simulation model source code, scripts and raw data are available at [[https://si-clusterraspberry.csr.unibo.it/downloads/Performance_evaluation_of_a_self-clustering_heuristic_for_adaptive_PADS.tgz|this link]]
-  * Raw data obtained by the experiments used to produce the figures: {{pads:results-paper-jpdc-2021.tgz|download}}