Distributed multiuser opportunistic spectrum access: A “win-stay loss-randomize” strategy

This paper studies the problem of distributed multiuser Opportunistic Spectrum Access based on Partially Observable Markov Decision Process (POMDP). Due to the similarity of spectrum environment, secondary users may choose the same channel adopting their own single user approach, which leads to collision. Referring to the previous works, we propose a more flexible and adaptive strategy named “win-stay loss-randomize“. If a secondary user succeeds in transmitting data and receives acknowledgment at one time slot, it still chooses the same channel as the current one at next slot or the slot after next according to different state transition probabilities Otherwise, secondary users continue selecting channels randomly. The simulation results shows that the upgrade of performance is up to 25% compared with the existing random strategy, which demonstrate the advantage of the proposed strategy.