Experimental evaluation of high-resolution direction-finding algorithms using a calibrated sensor array testbed

In this paper the performance of several direction-finding algorithms is investigated experimentally using the University of Minnesota Sensor Array Testbed. In recent years, many “high-resolution” direction-finding algorithms have been investigated using computer simulation and theoretical analysis. For the most part, no systematic experimental evaluation of these algorithms has been reported in the open literature. An experimental array testbed allows further evaluation of the capabilities and limitations of these algorithms. The system at the University of Minnesota basically consists of a linear array of eight ultrasonic transducers and several transmitters operating at 40 kHz in air. Nonideal conditions in actual arrays require calibration, and the methods of calibrating the sensor array testbed are described. Experimental results are presented, which compare the performance of several well-known algorithms, including Capon's MLM, MUSIC, Root-MUSIC, Min-Norm, ESPRIT, and a weighted norm version of MUSIC. The analysis includes experimental resolution thresholds, estimate bias and variance calculations, and comparisons with simulated results and the Cramer-Rao lower bound for variance.