A novel parallelized goodness-of-Fit test based Dynamic Spectrum access technique for Cognitive Radio Networks

Cognitive Radio Networks (CRNs) operate on the principle of opportunistically exploiting unused capacity in the primary network via Dynamic Spectrum Sensing. In this paper we propose a novel suite of transmit opportunity detection methods that effectively exploit the white/gray space in high traffic packet networks. We present a radically different paradigm to exploit the excess Signal-to-Noise ratio regime in which the primary network usually operates via a modification of the Interference Temperature concept. The proposed method is based on robust and rapid detection of changes in the primary network statistics through the use of a novel Parallelized Goodness-of- Fit test. The instantaneous transmit margin afforded by the primary network is dynamically determined and the CRN backs off whenever it detects that it is beginning to interfere with the primary network. We have implemented the proposed method on a CRN testbed that coexists with a large scale IEEE 802.11 primary network and demonstrate its excellent performance through extensive real world experimental results. We show that we can obtain more than 95% probability of detection of interference while the probability of not detecting a valid transmit opportunity is less than 20% for detection times of 400-1000 ms.