A software-based approach to active noise control in automobiles
The good news is that these measures help improve fuel economy; the bad news is that they often result in objectionable in-vehicle noise. This noise, or “boom”, exhibits one or more low-frequency tones under 150 Hz and can cause driver fatigue during long-term exposure. To counteract this noise, automakers are employing active noise control (ANC) technology.
How does ANC work?
ANC relies on a fairly simple acoustic principle. To cancel unwanted engine noise where the driver or passengers are sitting, an ANC system emits synthesized “anti-noise” of equal amplitude but opposite phase through the car speakers. That said, achieving optimal performance is anything but simple. To begin with, every vehicle interior has unique acoustic characteristics that are affected by the location of seats, the materials used in the cabin, and the position, number, and type of speakers and microphones. Because all of these factors influence how an ANC system performs, the system must be calibrated and tuned separately for every vehicle model. The system must also adapt quickly to dynamic changes in cabin acoustics that result from acceleration and deceleration, windows opening and closing, changes in seat positions, and fluctuations in temperature. Moreover, the system must be robust — it cannot become unstable or degrade the audio quality inside the cabin should, for instance, a microphone stop working.
To manage these changes effectively, an ANC system needs real-time engine operating data, specifically RPM data; it also needs input from one or more microphones located near the heads of the driver and passengers. The real-time engine data helps determine the frequencies to be reduced, while the microphones monitor noise levels in the cabin, enabling the system to adapt to the constant changes in the vehicle’s acoustic properties.
Figure 1 — ANC systems use real-time engine data and sampled microphone data to construct “anti-noise” that is played over the vehicle speakers to reduce engine noise where occupants are sitting.
Considerations and tradeoffs
Most commercial ANC systems use a dedicated hardware module. Automakers are beginning to realize, however, that it’s more cost effective to run software-based ANC on existing hardware, such as the head unit or power amplifier of the vehicle’s audio or infotainment system. Today’s infotainment systems must accommodate an increasing number of audio tasks such as hands-free processing and voice recognition, so the DSPs and application processors in these systems often have the headroom to run ANC as well.
Noise vibration and harshness engineers have the task of characterizing and optimizing sound quality in a vehicle. Choosing the number of microphones and speakers in a system is a trade-off between cost, performance, and computational resources, and often depends on vehicle size. Thus, an ANC solution should give automakers implementation choices. These include, for instance, the ability to run on a variety of processors or DSPs (with or without an OS) and the flexibility to accommodate almost any number and arrangement of inputs and outputs — from a one-microphone, two-speaker configuration to a six-microphone, six-speaker setup.
How does a software-based ANC solution measure up?
To answer this question, we conducted tests to compare noise reduction performance of the ANC software library from QNX Software Systems to that of an ANC system deployed in a popular 2012 minivan. The minivan uses cylinder deactivation to improve fuel consumption and has a dedicated, self-contained ANC module. The minivan system uses two microphones, one near the rearview mirror and the other in the center ceiling above the third row. It also uses up to four loudspeakers in the left and right front doors and sliding doors, and one subwoofer in the trunk.
Direct access to all of the existing microphones and speakers wasn’t feasible for the purposes of this test, so comparable microphones and speakers were placed as close as possible to the existing ones. One exception is the front microphone, which was placed closer to the driver’s head. Only three speakers were used for the software-based setup: left and right front-door speakers and the subwoofer.
Data was collected under multiple conditions, including a variety of cruising speeds on smooth quiet roadways, gradual and rapid acceleration to 50 mph, and throttling the engine with the vehicle in idle. As anticipated in a vehicle with cylinder deactivation, ANC had the greatest effect in the cruising speed conditions.
The software ANC system showed equal or better noise reduction than the production system. For instance, at a steady state speed of 65 mph, the production system demonstrated a 1 dB improvement with ANC activated. In comparison, the software ANC system showed a 12 dB improvement. Figure 2 illustrates the subset of the noise attenuation results at a cruising speed of 65 mph. Higher bars indicate better attenuation.
Figure 2 — Noise reduction measured in a minivan cruising at 65 mph shows how a software ANC solution can achieve up to 12 dB of noise reduction of the predominant engine frequency, outperforming a production system at all measured locations.
Enhancing the user experience
Consumer perception of overall vehicle quality hinges on interior sound quality. This includes the quality of hands-free calls and music playback, as well as an interior absent of engine and road noise. Automakers are introducing sound management technologies in different areas in the car, and next-generation vehicles will use technology that not only removes unwanted sound, but also injects desirable sounds into the car interior. For instance, engine sound enhancement technology generates customized sound synchronized to engine speed and emitted through the vehicle’s loudspeakers. Generated sound can also be piped externally to alert pedestrians of an approaching vehicle — especially beneficial with electric vehicles, which can be nearly silent. These new technologies will enhance the user experience, but uptake will depend on cost and on ease of integration within existing acoustic systems.
About the author :
Tina Jeffrey has more than 17 years’ experience in the semiconductor and software industries. She has designed SoCs and ASICs, and has expertise in both marketing and engineering strategy. Prior to joining QNX Software Systems as automotive product marketing manager, Tina was responsible for technical marketing of multimedia and image processors for automotive and consumer market segments. Tina has published papers on image cognition processors and intelligent imaging for automotive applications. She holds a Bachelor’s of Science degree in electrical engineering from the University of Ottawa.
For more information on QNX ANC technology, visit www.qnx.com or send email to anc@qnx.com.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
