The objective of the research is to develop analog spectrum processing techniques for cognitive radio (CR) applications in CMOS technology. CR systems aim to use the unoccupied spectrum allocations without any license when the primary users are not present. Therefore, the successful deployment of CR systems relies on their ability to accurately sense the spectrum usage status over a wide frequency range serving various wireless communication standards. Meanwhile, to maximize the utilization of the available spectrum segments, the bandwidth of the signal has to be highly flexible, so that even a small fraction of spectrum resources can be fully utilized by CR users. One of the key enabling technologies of variable bandwidth communication is a tunable baseband filter. In this research, a reconfigurable CR testbed system is presented as groundwork for the researches related with CR systems. With the feasibility study on the multi-resolution spectrum sensing (MRSS) functionality, a method for determining sensing threshold for MRSS functionality is presented, and a fully integrated MRSS receiver in CMOS technology is demonstrated. On the other hand, a reconfigurable CMOS analog baseband filter which can change its bandwidth, type and order with high resolution for CR applications is presented. In sum, an analog spectrum sensing method as well as a highly flexible analog baseband filter architecture is established and implemented in CMOS technology. Both designs are targeting the utilization of the analog signal processing capability with the aid of the digital circuits.