​During my industrial and academic career, I have faced various signal processing applications (electronic warfare, speech, radar, communications, physiological signals). This multiplicity was a true enrichment because it leads me to notice that prior knowledge of the signals encountered in practice is often too poor to justify the use of standard theoretical approaches. Motivated by this basic observation, I have been focusing on optimal statistical signal processing of signals in noise based on the least possible prior knowledge on the signals and their distributions. In this respect, I introduced the following theoretical frameworks that follow one after the other. Currently, my research mainly concerns the Random Distortion Testing (RDT) framework and the mathematical models of resilience, where RDT plays a crucial role.