Abstract—In order to increase Cognitive Radio operationefficiency, this paper builds up a new knowledge managementfunctional architecture for supporting spectrum management. Itintegrates the fittingness factor concept proposed by the authorsin a prior work and includes a set of advanced statistics capturingthe influence of the radio environment. Then, a KnowledgeManager (KM) exploiting these statistics and observed fittingnessfactor values has been developed to monitor the time-varyingsuitability of spectrum resources to support heterogeneous services.Based on estimated suitability levels, a new strategycombining Spectrum Selection (SS) and Spectrum Mobility (SM)functionalities has been proposed. Results have shown that theproposed strategy efficiently exploits the KM support at low loadsand the SM functionality at high loads to introduce significantgains (ranging from 85% to 100%) w.r.t. a pure random selectionwhile exhibiting substantial robustness to changes in interferencelevels.