HMM-based primary user activity in GSM via spectrum measurements

Primary user (PU) activity, as a key point in cognitive radio (CR) technologies and wireless communications, has been widely considered in various simulation environments, but still lacking the support of real-time spectrum data. In this paper, by measuring the power spectrum density (PSD) of GSM downlink in four consecutive days, we adopt a dynamic system based on hidden Markov model (HMM), which provides an ideal method to track the PU activity in GSM cellular mobile communication network. Generally speaking, the arrival of PU approximatively follows a Poissonian distribution, resulting in the exponential characteristic of traffic arrival time and traffic service time. By using the Baum-Welch (BW) algorithm in HMM, we can effectively estimate the parameters of PU behavior. Consequently, it is well-reasoned for us to acquire the state transition rate of PUs in different channels, as well as the sojourn time of each state. We demonstrate that the HMM-based system and the GSM cellular network match well in real-world spectrum environment.