Radar explores surrounding space by using a search pattern, ie a set of dwells. It must ensure detection at given range, using as low time budget as possible, and take into account anisotropic properties of the radar (deflection losses). Currently, search patterns are hand-designed. Our objectives are to provide: a framework for algorithmic generation of optimized search patterns adaptive scanning for inhomogeneous environments (clutter, ridges, ...) Methodology -Definition of the radar model on a discrete grid. -Formulation of the discrete model as an Integer Program -Optimization by Linear Relaxation and Branch&Bound