Addressing feasibility of cognitive radios

In this article we present a feasibility study of spectrum-sensing techniques using a research testbed platform for exploration and demonstration of cognitive radio systems. Our cognitive radio testbed is particularly suited for the development of physical and network layer functionalities and their experimental characterization in realistic wireless scenarios. Advanced testbed capabilities include real-time high-speed signal processing and protocol implementation, and support for multiple networks interaction and multiple antennas operation. This testbed is used for an experimental study of a set of prominent candidate techniques proposed in the literature for implementation of spectrum sensing functionality. We first consider three physical layer signal processing approaches based on energy, pilot, and feature detection. Our experimental results show that theoretical performance of these approaches for spectrum sensing cannot be achieved for the detection of very weak signals using a practical sensing receiver implementation. Also, we explain why these limitations exist by identifying sources of errors and provide basis for the design of robust spectrum sensing techniques. Next, we consider spectrum sensing techniques that use multiple measurements to improve sensing reliability. Our experimental study demonstrates practically achievable gains by exploiting multipath channel diversity through multiple antenna processing, and spatial diversity using user cooperation in a typical indoor environment.