Adversarial signal processing

Security-oriented applications of signal processing have received increasing attention in the last years. Digital watermarking, steganography and steganalysis, multimedia forensics, biometrics, intrusion detection, network monitoring, are just a few examples of such an interest. In many cases, though, researchers have failed to recognize the single most important feature behind any security-oriented application, i.e. the presence of one or more adversaries aiming at making the system fail. Even when the need to cope with the actions of a malevolent adversary is taken into account, the proposed solutions are often ad-hoc, failing to provide a unifying view of the challenges posed by such scenarios. Times are ripe to go beyond this limited view and lay the basis for a general theory that takes into account the impact that the presence of an adversary has on the design of effective signal processing tools, i.e. a theory of adversarial signal processing. It is the aim of this talk to: i) motivate the necessity of devising a general theory of adversarial signal processing, ii) describe how game theory can help to develop such a theory, iii) present some recent results regarding to adversarial hypothesis testing, iv) highlight some possible applications of adversarial signal processing to non-conventional fields like distributed detection and estimation, collaborative filtering, reputation systems.