Cognitive Radio Testbed: Exploiting Limited Feedback in Tomorrow's Wireless Communication Networks

Abstract

The next generation of wireless communication devices should support advanced features such as high spectral efficiency, broad bandwidth, diverse quality of service (QoS) requirements, and adaptivity. The cognitive radio (CR) is a new paradigm which has a high potential to become a basis for the future wireless systems. This paper is a first step towards the implementation of such a system. Our CR testbed is based on a GNU Radio platform which enables flexibility and reconfigurability of transmission parameters. As machine learning component, we invoke genetic algorithm (GA) to optimize the transmission parameters such as transmission power, modulation order and frequency channel based on the current spectrum conditions. Unlike other CR implementations, our approach requires very limited feedback information at the transmitter (ap 8 bits/packet duration). No transmission model nor additional network state information (NSI)is needed at the transmitter side. Experimentations show that our CR is capable to find free channels within 4-5 iterations even in a highly occupied spectrum scenario. It also offers the optimal trade-off between throughput, reliability, and power consumption depending on the user's QoS requirements.