Experimental Verification of CPU-GPU Architecture for ANC Algorithms

Active noise control (ANC) is a technology which lowers the noise level by using the principle of destructive interference of sound wave. Even though recent developments in digital signal processing (DSP) made it possible to implement ANC algorithms in real-time, its computational efficiency is not enough to handle various adaptive algorithms which show fast adaptation performances compared to traditional algorithms with LMS method or complex systems with multiple reference transducers, secondary speakers and error microphones. Unlike conventional DSP systems, Graphics Processing Unit(GPU) possesses massive computational power through parallel computation but is a co-processor which is connected to a processor via physical connection. In the previous research, CPU-GPU architecture for ANC algorithms is proposed so that massive computing power of GPU computation can fully be utilized to ANC algorithms while avoiding the negative effect of block data transfer between CPU and GPU memories. Throughout this research, verification of the previously proposed CPU-GPU architecture for ANC algorithms has been done through experiments.