9th International Conference on Cognitive Radio Oriented Wireless Networks

Research Article

Multi-Channel Selection Maximizing Throughput for Delay-Constrained Multi-Application Secondary Users in Dynamic Cognitive Radio Networks

Download540 downloads
  • @INPROCEEDINGS{10.4108/icst.crowncom.2014.255406,
        author={Luca Zappaterra and Hyeong-Ah Choi and Xiuzhen Cheng and Taieb Znati},
        title={Multi-Channel Selection Maximizing Throughput for Delay-Constrained Multi-Application Secondary Users in Dynamic Cognitive Radio Networks},
        proceedings={9th International Conference on Cognitive Radio Oriented Wireless Networks},
        publisher={IEEE},
        proceedings_a={CROWNCOM},
        year={2014},
        month={7},
        keywords={cognitive radio channel selection optimal stopping rule},
        doi={10.4108/icst.crowncom.2014.255406}
    }
    
  • Luca Zappaterra
    Hyeong-Ah Choi
    Xiuzhen Cheng
    Taieb Znati
    Year: 2014
    Multi-Channel Selection Maximizing Throughput for Delay-Constrained Multi-Application Secondary Users in Dynamic Cognitive Radio Networks
    CROWNCOM
    IEEE
    DOI: 10.4108/icst.crowncom.2014.255406
Luca Zappaterra1,*, Hyeong-Ah Choi1, Xiuzhen Cheng1, Taieb Znati2
  • 1: George Washington University
  • 2: University of Pittsburgh
*Contact email: lucaz@gwu.edu

Abstract

In dynamic cognitive radio networks (CRNs) secondary users (SUs) sense the spectrum bands to find temporal absence of primary users (PUs) and immediately transmit on the identified spectrum holes.
SUs sequentially sense the channels, stopping when the available resources are expected to provide the best throughput performance. Following, the selected channels are exploited using multi-channel transmission. In this paper, the multi-channel selection problem for SUs supporting multiple applications generating traffic with different latency requirements is formulated in a CRN with both heterogeneous PU-traffic and channel conditions. We have proposed an optimal solution overcoming the expensive computations and storage requirements typical of optimal stopping problems. Our efficient algorithm only requires linear-time and quadratic-space complexities in the online decision phase, aided by statistical decision values efficiently pre-computed offline. Extensive evaluations validate our solution as a significant improvement over the application of existing solutions, all based on the well-known backward induction technique or its approximations, characterized by either intractable algorithmic complexities or approximate results.