A planar antenna array based feature detection scheme is proposed to estimate the directional, location and modulation information pertaining to radio sources in a cognitive radio environment. The proposed system employs multiple direction estimation stations and a fusion station. Planar antenna arrays and three-dimensional (3-D) infinite impulse response (IIR) digital filters are employed to perform volume scanning of the radio environment, leading to a spatial power profile, which is subjected to peak detection in order to estimate the direction of arrival corresponding to each source. Cyclosationay feature detection is then performed along each direction to estimate the frequency and modulation information. Two simulation examples are provided to verify the feasibility of the proposed approach.