Article Preview
TopIntroduction
Noise emanating from mechanical, industrial and other sources have adverse effects on human health (Sierra et al., 2008). Noise pollution across various parts of the globe shows no sign of a slowdown, despite being its negative effects. Recent studies throw light into the unreasonable effects of environmental noise on human health, cognition, hearing ability, child development, etc. (Beutel et al., 2016). In the fMRI scanning process, the acoustic noise is often at high sound pressure levels causing discomfort to the patients as well as affecting the scanning results (Kannan, Milani, Panahi, & Briggs, 2011).
Active Noise Control (ANC) systems try to reduce the unwanted noise in specific regions of interest by generating an inverse acoustic signal (Lueg Paul, 1934). Figure 1 shows a block diagram of an ANC system. It has two acoustic channels, one from noise source to error microphone P(z) and another one from anti-noise source to error microphone Sa(z). The system uses W(z), an adaptive filter whose weights are updated iteratively according to the Least Mean Square (LMS) algorithm. Sb(z) represents the effects due to Digital to Analog Converter (DAC) and amplifier. The error signal is represented by e(n), x(n) is the reference signal, y(n) is the anti-noise signal, d(n) is the desired signal, and x^(n) is the filtered reference signal. The system employs an estimation of secondary path which is denoted by S^(z). ANC systems find wide applications in healthcare, automobile (Kajikawa, Gan, & Kuo, 2012), consumer electronics & mobile devices (Cheer, Elliott, Oh, & Jeong, 2018), the airline industry, etc.
Noise control mechanisms, both passive and active (Kuo, Mitra, & Gan, 2006) are in use for quite a while. Several decades of research have made progress in active noise control systems in areas including better algorithms, virtual sensing, and hearing aids (George & Panda, 2013). Still, the level of cancellation and the area of cancellation for acoustical quite zones are open problems for the ANC research community (Zhang, Abhayapala, Zhang, Samarasinghe, & Jiang, 2018).
ANC systems find its use case application in small (e.g. fMRI room or airline cabin) to large rooms (e.g. industrial cases). Each room is considered to have its own unique properties like reverberation time and resonant frequencies. Characteristics of a room in ANC system design are normally not considered for simplicity.
Figure 1.
Block diagram of the ANC system. P(z) is the primary acoustic channel and Sa(z) is the secondary acoustic channel.
Simultaneous multi-channel RIRs were successfully used to reconstruct room geometry (Dokmanic, Parhizkar, Walther, Lu, & Vetterli, 2013) (Jager, Heusdens, & Gaubitch, 2016). It has been shown that noise source localization in ANC by employing room acoustics improves the result by 5dB (Liu, Li, & Kuo, 2018) (Hase, Kajikawa, Liu, & Kuo, 2015). Similar applications like room equalization have also found successful use cases. In fMRI rooms, the equipment and patient positions are normally fixed. Also, each fMRI room is likely to have its own geometrical shape. This brings up the possibility to improve ANC by making good use of the room and acoustic channel properties.