The present invention relates to a blind detection of a received signal having at least one property, which initially is unknown to the receiver, however, limited to a finite set of alternatives ({p.sub.1, p.sub.2, . . . , p.sub.m}). According to the invention, the unknown property is automatically identified as follows. A respective quality measure (q.sub.1, q.sub.2, . . . , q.sub.m) is derived from the incoming signal (r(n)), which each represents a particular property in the finite set of alternatives ({p.sub.1, p.sub.2, . . . , p.sub.m}). The quality measures (q.sub.1, q.sub.2, . . . , q.sub.m) are obtained according to a procedure that involves a computationally efficient rejection of any unwanted signal components in the incoming signal (r(n)). Then, the quality measures (q.sub.1, q.sub.2, . . . , q.sub.m) are compared with each other in order to find a quality measure, which best fulfills a blind selection criterion. The quality measure (q.sub.i) thus obtained corresponds to the initially unknown property (p.sub.i). For instance, based on this information, the incoming signal (r(n)) may then be further processed in a processing unit (122), which operates according to a processing principle (C.sub.di) that is adapted to this property (p.sub.i).