Investigation of high-frequency propagation channels through pipes and ducts for building interior reconnaissance

Recently, there is strong interest in the through-wall sensing capabilities of radar for use in law enforcement, search and rescue, and urban military operations. Due to the high attenuation of walls, through-wall radar typically operates in the low GHz frequency region, where resolution is limited. It is worthwhile to explore other means of propagating radar waves into and back out of a building’s interior for sensing applications. One possibility is through duct-like structures that are commonly found in a building, such as metal pipes used for plumbing or air conditioning ducts. The objective of this dissertation is to investigate techniques to acquire radar images of targets through a pipe. First, using the pipe as an electromagnetic propagation channel is studied. A modal approach previously developed for computing the radar cross-section of a circular duct is modified to compute the transmission through a pipe. This modal approach for transmission is validated against measured data. It is also shown that a pipe is a high-pass propagation channel. The modal analysis is then extended to two-way, through-pipe propagation for backscattering analysis. The backscattering from a target is observed through a pipe in simulation and measurement. Next, methods to form two-dimensional radar images from backscattering data collected through a pipe are explored. Four different methods previously developed for free-space imaging are applied to the problem of imaging through a pipe: beamforming, matched filter processing, MUSIC, and compressed sensing. In all four methods it is necessary to take into account the propagation through the pipe in order to properly generate a focused radar image. Each method is demonstrated using simulation and validated against measurement data. The beamforming and matched filter methods are found to suffer from poor cross-range resolution. To improve resolution, the MUSIC algorithm is applied and shown to give superior resolution at the expense of more complicated data collection. The final method, compressed sensing, is shown to achieve good cross-range resolution with simpler data collection. A comparison of the tradeoffs between the four methods is summarized and discussed. Two additional extensions are studied. First, a method for computing the transmission through an arbitrary pipe network using the generalized scattering matrix approach is proposed and implemented. Second, a new method for computing joint time-frequency distributions based on compressed sensing is applied to analyze the backscattering phenomenology from a pipe.