Intelligent Ultra Low-Power Signal Processing for Automotive

The acoustic sensing and processing cases on smart cars form a complicated implementation scenario, including state-of-the-art sensor geometries, learning-based algorithms, and customized processor architectures. In this thesis, the whole workflow is integrated into a single system with an efficient and low-power chip as the dedicated hardware processor. This workload-adaptive and ultra-low-power processor will enable a low-latency and always-active response for acoustic reasoning algorithms.