Reducing HLA Time Management Overhead via Preemptive Execution

Distributed co-simulation is widely used to study 5G/6G systems by coupling cellular network simulators with traffic and mobility application models using the IEEE 1516 High-Level Architecture (HLA). When one federate (e.g. a network simulator) generates thousands of fine-grained events per second with zero lookahead, and another (e.g. a traffic simulator) advances in coarse, fixed time steps, the resulting asymmetry incurs severe HLA time-management overhead. In particular when simulating cellular networks, every millisecond a new frame triggers a Next-Event-Request (NER) and Time-Advance-Grant (TAG) exchange with the Runtime Infrastructure (RTI), dramatically slowing overall progress. We propose a conservative, preemptive execution scheme. This simple modification preserves sequential time management, avoids optimistic rollback, and - depending on the network load - reduces wall-clock runtime by 25-50%. The method directly benefits network simulation and modeling, as well as large-scale studies of 5G/6G systems where heterogeneous time scales are unavoidable.