Case-Based Cognitive Cellular Systems for Temporary Events

Abstract—In this study we investigate the use of case-based reinforcement learning (RL) for dynamic secondary spectrum sharing in cognitive cellular systems for temporary events. The performance of the proposed case-based RL scheme is evaluated using system level simulations that involve a stadium small cell network, an eNB on an aerial platform and a local primary LTE network. Compared to classical RL, the case-based RL approach results in increased adaptivity of the cognitive cellular system to sudden changes in its environment caused by the aerial eNB being dynamically switched on and off. We also show that a cognitive cellular system that employs the proposed dynamic spectrum access scheme is able to accommodate a 51-fold increase in offered traffic with no need for additional spectrum and with no degradation in the quality of service of the primary users