Predictive behavior classification for cognitive radio: Introduction and preliminary results

Cognitive Radio systems rely heavily on artificial intelligence capabilities to perform a variety of tasks. Sharing spectrum resources more efficiently, self organization, and interference mitigation are just a few examples. For many CR applications, a primary goal is to decentralize and distribute network functions among participant nodes. As a consequence, any given node in a CR network may be required to coordinate with not only its peers, but also with a number of unknown transmitters. Thus, it is desirable that individual nodes be capable of predicting future states of non-peer transmitters in order to better optimize their own operation. In this paper we introduce methods for identifying cognitive behavior in an unknown transmitter and predicting likely future states based on physical spectrum observations. We discuss the problem in the context of our Universal DSA Network Simulation (UDNS) and present two behavior classification algorithms used to this end.

[1]  Nir Friedman,et al.  Bayesian Network Classifiers , 1997, Machine Learning.

[2]  János Abonyi,et al.  Computational Intelligence in Data Mining , 2005, Informatica.

[3]  Geoffrey I. Webb,et al.  Averaged One-Dependence Estimators: Preliminary Results , 2002, AusDM.

[4]  Michael I. Jordan,et al.  Unsupervised Learning from Dyadic Data , 1998 .

[5]  Tamal Bose,et al.  Classification of DSA channel selection for estimation of secondary user behavior , 2011, 2011 IEEE 54th International Midwest Symposium on Circuits and Systems (MWSCAS).

[6]  Jose Miguel Puerta,et al.  GAODE and HAODE: two proposals based on AODE to deal with continuous variables , 2009, ICML '09.

[7]  Andrew W. Moore,et al.  Internet traffic classification using bayesian analysis techniques , 2005, SIGMETRICS '05.

[8]  Yang Yu,et al.  Semi-naive Exploitation of One-Dependence Estimators , 2009, 2009 Ninth IEEE International Conference on Data Mining.