Dynamic Spectrum Tracking Using Energy and Cyclostationarity-Based Multi-Variate Non-Parametric Quickest Detection for Cognitive Radios

A novel non-parametric, multi-variate quickest detection method is proposed for cognitive radios (CRs) using both energy and cyclostationary features. The proposed approach can be used to track state dynamics of communication channels. This capability can be useful for both dynamic spectrum sharing (DSS) and future CRs, as in practice, centralized channel synchronization is unrealistic and the prior information of the statistics of channel usage is, in general, hard to obtain. The proposed multi-variate non-parametric average sample power and cyclostationarity-based quickest detection scheme is shown to achieve better performance compared to traditional energy-based schemes. We also develop a parallel on-line quickest detection/off-line change-point detection algorithm to achieve self-awareness of detection delays and false alarms for future automation. Compared to traditional energy-based quickest detection schemes, the proposed multi-variate non-parametric quickest detection scheme has comparable computational complexity. The simulated performance shows improvements in terms of small detection delays and significantly higher percentage of spectrum utilization.