IntroductionRF tomography has been proposed to localize and image man-made underground structures and tunnels using a set of arbitrarily distributed narrowband electromagnetic sensors placed above or below the ground. Basically, a set of s transmitters emit a monochromatic signal into the ground; waves impinge upon dielectric anomalies (e.g. tunnels) thus generating a scattered field. Finally, distributed receivers sample the scattered field and relay this information to a base station for post processing. For a deeper understanding of the principles of RF tomography for belowground imaging, see [1]. Currently, RF tomography is based upon the knowledge of the Green function of the scattering problem: analytic expressions are limited only to layered media with planar interfaces [5]. For irregular terrains, a numerical Green function has to be evaluated and here we show a proper reformulation of RF tomography that accounts for numerical Green function that describes the spatial impulse response of the terrain shape. This improvement lead to more reliable reconstructions compared to the case of a planar interface.
Figure 1: Geometry
RF Tomography Assuming Flat SurfaceWe consider the 3D transmitter-receiver geometry depicted in Fig. 1. The n-th electrically small dipole (length l ) acting as transmitter is located at position a n r , with moment direction ˆn a and current intensity n I , and the m-th electrically small dipole acting as receiver is located at position b m r , with moment direction ˆm b . Each transmitter emits a monochromatic signal with frequency f . We first assume the air-earth interface to be flat: in the next section, we will address the problem of non-flat interface. The air half-space is modeled as free-space medium, while the ground half-space is modeled as an homogeneous medium with background relative dielectric permittivity D ε , 978-1-4244-4968-2/10/$25.00 ©2010 IEEE background conductivity D