Predictive decision-making for vehicular cognitive radio networks through Hidden Markov models

Vehicular networks that require additional spectrum for communication may leverage cognitive radio technology. Towards this aim, the vehicle must select one among several candidate channels for data transmission, with the possibility that other surrounding vehicles may also identify the same spectrum for their use. Such networks are distinguished from their classical, stationary counterparts by high mobility, time-varying and heterogeneous environment leading to dynamic channel availability. Owing to this changing environment, the history of spectrum usage information may be unavailable to the vehicles at all points on their journey, necessitating blind decision-making. In this paper, we propose a method of spectrum selection without a priori channel information. The main contributions of our work are: (i) we provide a framework to determine, through a Hidden Markov Model, whether a channel is occupied by a licensed user, another cognitive radio enabled vehicle, or if the observed signal fluctuations are due to noise; and (ii) we devise a prediction algorithm to determine which channels are likely to be available within the shortest amount of time. Our approach is verified through a simulation study, and we demonstrate equivalent performance to schemes that use prior history of the channel usage at specific vehicle locations.