Cognitive radio draws lots of research attentions in recent years for its efficient spectrum utilization. This research proposes a discrete wavelet packet transform (DWPT)-based spectrum sensing processor for cognitive radio systems. In this processor, this study proposes a partial DWPT spectrum analyzer and a double-threshold energy detection and decision result prediction technique to reduce the computational complexity. Complexity analysis shows that the proposed spectrum sensing processor saves at most 62% computational cost compared to the traditional FFT processor. VLSI implementation results also show that the proposed architecture saves about 40% energy dissipation for the IEEE 802.22 WRAN system.