Technological contributions to imaging radars in the millimeter-wave band

Resumo

Recent research in the millimeter- and submillimeter-wave frequency bands has driven the maturity of its technologies to a level where integrated and commercial radiofrequency components are now readily available. Due to this development the fundamental applications of these frequency bands (like spectroscopy, radioastronomy, and wireless communication) have benefited from increasing performance, at a fraction of the cost. This is also the case for imaging applications, such as concealed weapon detection, which is progressively becoming a viable alternative to traditional security screenings. Nonetheless, some technological progress is still needed for security imaging in this frequency range to fully reach its potential, specially regarding critical and diverse areas such as component cost, power consumption, imaging speed, or privacy concerns. The main goal of this Ph.D. thesis can be found within this framework, and it is to develop those technological improvements and solutions necessary to enhance the performance of security imaging radars in the millimeter- and submillimeter-wave bands. This objective has been accomplished in two particular radar systems: an active imager at 300 GHz (developed previously to the start of this thesis), and a hybrid system using active and passive sensors at 122 GHz and 94 GHz, respectively. Throughout the development, a special emphasis is placed on achieving the desired results with the use of commercially-available components, in an effort to truly provide a viable alternative. Although both systems are meant for security screening purposes, their radiofrequency and processing architectures present various differences according to their operational requirements. In addition to their security-related function, both systems have also been employed in other applications, such as non-destructive testing or vital sign monitoring. In the 300 GHz radar, the developments of this thesis have enabled a real-time image processing architecture achieving 9.42 fps image refresh. In addition, a new multistatic imaging configuration for this radar based on multiple bistatic measurements has been designed and tested. For the hybrid system, an active 122 GHz sensor has been developed through the use of a commercial MMIC, and an image fusion algorithm has been designed to combine the imaging results from different classes of sensors. Deep learning models for automatic threat detection have also been developed and tested during this thesis, achieving accuracies higher than 99% for both active and passive images. Furthermore, the 300 GHz and 122 GHz have also demonstrated great performance in several alternative applications aside from security screening, such as the detection and registration of a human's respiration and heart beat at 8 m distance. The thesis has been entirely carried out at the Grupo de Microondas y Radar of the Universidad Politécnica de Madrid, and is divided into four main parts. The first section deals with the improvements made to the 300 GHz imaging system, including a real-time signal and image processing architecture and the initial tests of a multistatic configuration. The second part is focused on the development of the hybrid imager at 122 GHz and 94 GHz, detailing individual sensor characterizations, simultaneous performance, and an image fusion technique to present the captured data. The third section describes the development of the automatic threat detection models for active and passive images. Lastly, the fourth section summarizes the use of the radars in alternative applications. Additionally, pending improvement actions are also studied and described for future research and work.