Application of MUSIC to microwave imaging for detection of dielectric anomalies

We apply the non-iterative MUltiple SIgnal Classification (MUSIC) algorithm to identifying the locations/shapes of dielectric anomalies from collected scattered field S-parameters. By identifying the mathematical representation between MUSIC and an infinite series of Bessel functions of integer order of the first kind when the total number of antennas is small, we examine some properties of MUSIC in microwave imaging. Simulation results obtained from synthetic data at f = 1 GHz frequency using CST studio demonstrate the effectiveness and limitations of MUSIC.

1. Introduction

In this paper, we study the microwave imaging of a single, small homogeneous anomaly located in homogeneous space. This anomaly is characterized by the contrast between dielectric permittivity and electrical conductivity with respect to the space. The shape or location of such anomalies has been studied previously using iterative schemes or level set methods. Related works can be found in [1]–[6] and the references therein. Unfortunately, a number of parameters need to be fulfilled to proceed with an iterative scheme: high computational costs, selecting an appropriate regularization that depends upon the problems at hand, calculation of complex Fréchet derivatives, and a good initial shape or location of the anomaly. Therefore, it is very hard to apply such a scheme to various practical situations, so fast noniterative schemes have been developed as an alternative.