Radio Frequency (RF) tomography has been proposed for imaging dielectric and conducting anomalies above-ground. Accordingly, low-cost electromagnetic transmitters are placed arbitrarily above ground, surrounding
a large area of interest. In a preliminary stage, sensors identify their position, orientation, and time reference.
Subsequently, a transmitter radiates a known waveform. The probing wave impinges upon a target (represented
in terms of dielectric or conducting anomaly), thus producing scattered elds. Spatially distributed receivers
collect samples of the total electric eld, remove noise, clutter and the direct path, and store the information
concerning only the scattered eld. In the next iteration, a dierent transmitter is activated, or dierent wave-
forms are used. Then, the collected data is typically relayed to a centralized location for processing and imaging.
To ensure persistent sensing, fast back-propagation algorithms are implemented (either involving correlation or
multiplication by a hermitian matrix). Resolution using back-propagation is aected by the sidelobe structure
of the ambiguity function of the wave. Clearly, Linearly Stepped Frequency (LSF) waveform requires the lowest
instantaneous bandwidth, but produces poor correlation properties. On the converse, Noise waveforms exhibit
the idealized thumb-tack ambiguity function but typically require large instantaneous bandwidths. In an eort
to exploit the benets of both individual waveforms, a noisy LSF waveform is developed. The NLSF performance,
limitation and spectral dominance in reference to RF Tomography, along with its theoretical bounds, will be
provided. Reconstructed images from simulated and experimental data will be compared.