Increasing resolution is an attractive goal for all types of radar sensor applications. Obtaining high radar resolution is strongly related to the signal bandwidth which can be used. The currently available frequency bands however, restrict the available bandwidth and consequently the achievable range resolution. As nowadays more sensors become available e.g. on automotive platforms, methods of combining sensor information stemming from sensors operating in different and not necessarily overlapping frequency bands are of concern. It will be shown that it is possible to derive benefit from perceiving the same radar scenery with two or more sensors in distinct frequency bands. Beyond ordinary sensor fusion methods, radar information can be combined more effectively if one compensates for the lack of mutual coherence, thus taking advantage of phase information.

At high frequencies, complex scatterers can be approximately modeled as a group of single scattering centers with constant delay and slowly varying amplitude, i.e. a set of complex exponentials buried in noise. The eigenanalysis algorithms are well known for their capability to better resolve complex exponentials as compared to the classical spectral analysis methods. These methods exploit the statistical properties of those signals to estimate their frequencies. Here, two main approaches to extend the statistical analysis for the case of data collected at two different subbands are presented. One method relies on the use of the band gap information (and therefore, coherent data collection is needed) and achieves an increased resolution capability compared with the single-band case. On the other hand, the second approach does not use the band gap information and represents a robust way to process radar data collected with incoherent sensors. Combining the information obtained with these two approaches a robust estimator of the target locations with increased resolution can be built.