Interactive Acoustic Simulation in Urban and Complex Environments

Outdoor acoustics simulation plays a vital role in several army and defense-related applications, such as minimizing the noise profile of reconnaissance vehicles to avoid counter-detection, optimizing sensors and systems for tactical advantage, and pinpointing the origin of gunshots using their acoustic signatures. Outdoor acoustics prediction technology is needed to efficiently model large, complex, and dynamic urban and battlefield scenes. State-of-the-art outdoor acoustics prediction methods are restricted, slow, or limited to simple environments. Specifically, no existing method can simultaneously model atmospheric acoustical phenomena, surface interactions with terrain and obstacles, and moving sound sources. We propose to develop a novel hybrid acoustics prediction technique, called multi-domain acoustic transfer (MDAT), which combines numerical and geometric acoustics algorithms. It would handle complex urban and outdoor environments, while running on commodity hardware. It would model broad frequency ranges, atmospheric effects, surface interactions, and dynamic environments with moving sound sources and receivers. Phase I would involve developing the algorithm and a prototype implementation, and evaluating its performance on simple scenes. Phase II would involve developing a full implementation capable of handling large-scale dynamic environments, exploiting multi-core CPUs and many-core GPUs. This implementation would be integrable into existing Army and DoD systems in Phase III.