Machine Learning and Intelligent Communications. Third International Conference, MLICOM 2018, Hangzhou, China, July 6-8, 2018, Proceedings

Research Article

Dynamic Opportunistic Spectrum Access with Channel Bonding in Mesh Networks: A Game-Theoretic Approach

Download
138 downloads
  • @INPROCEEDINGS{10.1007/978-3-030-00557-3_38,
        author={Chen Pan and Yunpeng Cheng and Zhengju Yang and Yuli Zhang},
        title={Dynamic Opportunistic Spectrum Access with Channel Bonding in Mesh Networks: A Game-Theoretic Approach},
        proceedings={Machine Learning and Intelligent Communications. Third International Conference, MLICOM 2018, Hangzhou, China, July 6-8, 2018, Proceedings},
        proceedings_a={MLICOM},
        year={2018},
        month={10},
        keywords={Opportunistic spectrum access Dynamic users Channel bonding Potential game},
        doi={10.1007/978-3-030-00557-3_38}
    }
    
  • Chen Pan
    Yunpeng Cheng
    Zhengju Yang
    Yuli Zhang
    Year: 2018
    Dynamic Opportunistic Spectrum Access with Channel Bonding in Mesh Networks: A Game-Theoretic Approach
    MLICOM
    Springer
    DOI: 10.1007/978-3-030-00557-3_38
Chen Pan1,*, Yunpeng Cheng1,*, Zhengju Yang2,*, Yuli Zhang1,*
  • 1: Army Engineering University of PLA
  • 2: No. 92274 Troops of PLA
*Contact email: panchenlgdx@163.com, ypcheng@yahoo.com, yangzhengju1001@126.com, yulipkueecs08@126.com

Abstract

The opportunistic spectrum access with dynamic users and channel bonding technology in mesh networks is studied in this paper. Different from the traditional static and fixed transmitting model, nodes would change their states between active and silent, due to their traffic demand. Also, the channel bonding technology, which mitigates interference and improves throughput significantly, is employed in this paper. The interference mitigation problem with channel bonding is modeled as a distributed and non-cooperative game. We proved it to be an exact potential game. Based on the good property of the potential game, it guarantees the existence of at least one pure Nash equilibrium (NE). Due to the potential function is formulated as the aggregate interference of the network, the final optimal NE point also achieves the minimization of the system’s total interference. A multiple-agent learning algorithm is designed to approach the NE points. Compared with other algorithms, simulation results show that the modified algorithm achieves a lower interference performance, and the channel bonding contributes to the throughput performance.