i. Parallelization: ns-3 simulations execute sequentially on a single CPU core, the ability to parallelize wireless simulations has been somewhat of an open research question. Some promising recent work from our collaborators at Lawrence Livermore National Laboratory (LLNL) opens the door to realizing effective parallelization for wireless simulations within ns-3. While LLNL is focusing on parallelizing 5G cellular scenarios, we will focus on the same for Wi-Fi and plans to transfer techniques used in ns-3 to simulate cell-edge interference to reduce the event coupling between parallel wireless simulation processes.
ii. Improved Link Abstractions: Achieving run-time gains by parallelizing must be complemented by more efficient execution of simulation models in each (parallelized) process. To this end, we will pursue a few approaches: a. identify and prune states corresponding to wireless interference events that have little or no bearing on packet reception outcomes, and b. seek novel link-to-system mappings for complex transmis- sion scenarios (fading, MIMO and multi-user channels) that can represent such inherently computationally burdensome use-cases much more efficiently while preserving simulation fidelity.
Georgia Tech's activities center on high-fidelity mmWave channel models that incorporate environmental effects, i.e. blockages and reflections that occur from obstacles such as walls, buildings, furniture, and even human beings. Georgia Tech is also working on optimistic parallel simulations for limited-interference settings such as occur with highly-directional antennas in the mmWave bands. Georgia Tech is also working on educational materials to help new ns-3 users work with the advanced wireless features in the simulator.