Neural network-based learning schemes for cognitive radio systems

Intelligence is needed to keep up with the rapid evolution of wireless communications, especially in terms of managing and allocating the scarce, radio spectrum in the highly varying and disparate modern environments. Cognitive radio systems promise to handle this situation by utilizing intelligent software packages that enrich their transceiver with radio-awareness, adaptability and capability to learn. A cognitive radio system participates in a continuous process, the ''cognition cycle'', during which it adjusts its operating parameters, observes the results and, eventually takes actions, that is to say, decides to operate in a specific radio configuration (i.e., radio access technology, carrier frequency, modulation type, etc.) expecting to move the radio toward some optimized operational state. In such a process, learning mechanisms that are capable of exploiting measurements sensed from the environment, gathered experience and stored knowledge, are judged as rather beneficial for guiding decisions and actions. Framed within this statement, this paper introduces and evaluates learning schemes that are based on artificial neural networks and can be used for predicting the capabilities (e.g. data rate) that can be achieved by a specific radio configuration. In particular, the focus in this work is placed on obtaining insight on the behavior of the presented, learning schemes, whereas useful, indicative results from the benchmarking work, conducted in order to design and use an appropriate neural network structure, are also presented and discussed. In the near future, such learning schemes are expected to assist a cognitive radio system to compare among the whole of available, candidate radio configurations and finally select the best one to operate in.