The Helmut Schmidt University/University of the Federal Armed Forces Hamburg continues to work at full speed on systems for active noise reduction by means of anti-noise. This includes measures against noise pollution in apartments caused by constantly increasing traffic. dSPACE Magazine presented promising results in 2014. New technologies and options of the new SCALEXIO real-time system helped researchers make a significant leap in the development of their solution and simplify its everyday use.
The omnipresent noise pollution, especially in urban areas, that is mainly caused by traffic creates stress and damages people’s health. However, the insulating materials that are currently being used in residential buildings lose their effectiveness if the sound frequency decreases, as is the case for truck engines, for example. In addition, even a window tilted for ventilation can ruin even the best passive insulation. This is exactly where the concept of active noise reduction comes in.
Figure 1: The principle of active noise reduction: The original sound wave (gray) and the canceling sound wave (blue) are superimposed and (almost) cancel each other out.
Active sound reduction is based on the principle of destructive interference, in which two opposite-phase waves cancel each other out. The anti-noise is calculated on the basis of the measurements taken by two microphones (Figure 1) and superimposed with the interference signal via a loudspeaker. The first microphone near the source of the noise detects the unwanted noise, while the second microphone is located in a quiet area measuring the resulting error signal from the superposition of noise and anti-noise. In practice, the noise cannot be canceled fully due to various interference factors. Sound propagates in all directions and is also reflected by surfaces. This results in a complex sound field with a broad frequency spectrum for which at best only a very local anti-noise signal can be generated.
Figure 2: Structure of the anti-noise test system including dSPACE components.
The setup shown in Figure 2 consists of two rooms: a low-reflection outer room with loudspeakers for sound creation, and an inner room whose acoustic properties are typical of a room in a residential building. The inner room is connected to the outer room by a standard window.
The generation of the anti-noise is based on a multichannel system that uses an internally calculated reference signal based on the measured error for each channel instead of the reference microphone illustrated in Figure 1. This simplifies the system design. However, this simplification requires extremely short calculation times, because the anti-noise must be calculated and generated on the basis of the predicted noise before the noise reaches the anti-noise loudspeaker. The shorter the period of time for this prediction, the better the noise reduction.
The signal processing was implemented on the basis of a SCALEXIO real-time system. The analog signals of the 16 error microphones used are routed via two DS2655 FPGA Base Boards, each equipped with 5 DS2655M1 Multi-I/O Modules, to two DS6001 Processor Boards that calculate the output signals for the anti-noise with an adaptive digital control algorithm (filtered-x-least-mean-square). These signals are then converted from digital to analog and output in an amplified form via 16 anti-noise loudspeakers.
Figure 3: For the perfect night's sleep: Experimental setup with loudspeakers and microphones integrated into the window frame.
All areas of the system have been revised and improved since 2014. For example, the microphones and loudspeakers are now placed inside the window frame. Therefore, none of the components are placed outside the building and exposed to the weather. In addition, the view through the window is not obstructed by the additional components. Thus, a retrofit kit can be designed to be installed to existing windows, if required.
By using the new SCALEXIO real-time system with its significantly improved processor performance and system bandwidth compared to the previously used PHS technology, the team was able to significantly improve signal processing. The sampling rate of the anti-noise system was increased by a factor of five to 40 kHz while also increasing the number of channels. This means that additional analog low-pass filters in the signal path were no longer required. The performance of the SCALEXIO real-time system is illustrated by the following successful configuration of the test setup:
The current structure is primarily aimed at reducing static noise sources, such as construction noise. However, especially moving noise sources, such as passing trucks or approaching buses on busy roads, are a major issue.
In the next phase of the project, these noise sources will be specifically investigated and countermeasures will be explored. The unutilized potential of the dSPACE system will also be investigated further. For example, the complexity of the calculations can be further increased by interconnecting the two processor systems to form a multiprocessor system and by outsourcing signal preprocessing to the FPGA.
Hptm M. Sc. Jonas Hanselka works in the in the Mechatronics department of the Helmut Schmidt University/University of the Federal Armed Forces Hamburg.
Subscribe to our newsletters, or manage or delete your subscriptions