Performance Analysis of the Probabilistic Models of ISM Data Traffic in Cognitive Radio Enabled Radio Environments

Abstract

Cognitive radio (CR) provides a way to utilize the radio resources in a smart manner. In order to exploit the benefits of CR technology either in CR systems or in CR enabled wireless sensor networks, CR based smart grids, CR based wireless body area networks (WBANs) and CR based Internet of Things (IOTs), it is necessary to characterize the radio traffic in a realistic way. In perspective of CR operating models, the radio traffic can be modeled either using conditional or unconditional modeling approaches. In unconditional modeling approach, observed data traffic is solely considered as interference while in conditional modeling approach the observed data traffic is first classified as signal and noise. Prior knowledge about the statistics of interference in unconditional modeling approach while signal (primary user) and noise (secondary user) characterization in conditional modeling framework will help in efficient utilization of radio resources within CR-enabled radio environment. Furthermore, performance analysis is carried out for both the unconditional and conditional models based on devised information theoretic criterion. Multivariate Gaussian mixture (MGM) as a special case of hidden markov random field (HMRF) and Gaussian mixture (GM) are suitable candidate models for the observed data traffic in ISM (Industrial, Scientific and Medical) band based on devised criterion in unconditional and conditional models, respectively.