Improving Radio Tomographic Images using multipath signals

Radio Tomographic Imaging (RTI) seeks to reconstruct useful images of the environment through measurements of received signal strength (RSS) in a network of radios. RSS data is compiled over many transmissions, and after processing the data an estimated image of the spatial loss field (SLF) of the area can be recovered. This image represents the estimated radio frequency signal attenuation in dB that occurs at each pixel location. An ideal RTI system could form images of floor plans and objects that are located inside buildings or behind other obstructions [1]. One of the challenges in RTI is the presence of multipath signals. Such signals reach the receiver at some delay with respect to the line-of-sight (LOS) signal. Presently, there has been no attempt to deal with such signals other than to reject them, at best. Thus, potentially useful links in a network of radios cannot be utilized, because their LOS signal path does not pass through the area to be imaged. This paper presents a method by which multipath signals may be treated as useful measurements in an RTI system. Doing so informs the existing SLF image, improving image quality and reducing root-mean-squared-error (RMSE), which is the focus of this paper. It is also possible to use such measurements to create an estimate of the area's reflectivity loss field, or RLF. Such an estimated image represents the reflection coefficient (in dB) seen at each reflective pixel location. This image could be useful in identifying interior features that are reflective, such as large metal objects or metal walls. Thus, two separate images are estimated simultaneously, an SLF image of pixel attenuation, and an RLF image of pixel reflectivity values. Both estimated images remain linear functions of the received data. We refer to this new model as the Augmented Multipath Linear Generating Model, or AMLGM. After presenting the AMLGM model, we discuss the results of our simulations showing improvements in both subjective and objective image quality.