Online Learning Anti-Jamming Cognitive Radio Network for Green Clouds Recently, the cognitive radio network (CRN) has been proposed as a candidate communication technology for cloud systems due to the flexibility provided by the CRN. The CRN employs opportunistic spectrum access (OSA) techniques to reuse the spectrums licensed to the primary users (PUs). However, most current OSA schemes do not consider jamming attacks against the CRN nodes. In this paper, we model the problem of anti-jamming dynamic multichannel sensing and access as a multi-armed bandit (MAB) problem, where the transmitter and the receiver adaptively choose arms (i.e., sending and receiving channels) to operate based on the common feedback. We apply two modified online machine learning algorithms to the cognitive nodes based on Markov chains. Our simulation results show that the proposed schemes are resilient to various jamming attacks.