ASA 130th Meeting - St. Louis, MO - 1995 Nov 27 .. Dec 01

3aEA15. Reference sensor selection for automotive active noise control applications.

Kevin D. LePage

P. J. Remington

Alan R. D. Curtis

Bolt Beranek and Newman Inc., 70 Fawcett St., Cambridge, MA 02138

Scott Knight

Bolt Beranek and Newman Inc., Union Station, New London, CT 06320

BBN has obtained a comprehensive set of road noise measurements for the purpose of investigating the potential performance of an automotive active noise control application. Measurements were collected over a total of 20 reference sensors on or near the suspension and four residual sensors in the cabin. Analysis shows that while the reference sensors can account for much of what is received at the microphones, some of the reference sensors provide more residual information than others. For control applications the optimum reference sensors for the available number of input channels are desired. Two reference sensor performance metrics are compared. The first is a suboptimal greedy algorithm where reference channels are iteratively selected via an orthogonalization procedure. The second technique is an exhaustive search over candidate reference sensors sets where the coherent power is maximized. Results show that the greedy procedure generally identifies the majority of the sensors shown to be optimum through the exhaustive search. [Work supported by ARPA.]