A Resource Analysis of Radio Space Distribution for the Wide-Area Virtualization of Wireless Transceivers


IEICE TRANSACTIONS on Communications   Vol.E97-B    No.9    pp.1800-1807
Publication Date: 2014/09/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E97.B.1800
Type of Manuscript: Special Section PAPER (Special Section on Ambient Intelligence and Sensor Networks)
ubiquitous network,  wireless transceivers,  virtualization,  sensing platform,  

Full Text: PDF>>
Buy this Article

Wide area virtualization of wireless transceivers by centrally managed software radio systems is a way to efficiently share the resources for supporting a variety of wireless protocols. In order to enable wide-area virtualization of wireless transceivers, the authors have developed a mechanism to deliver the radio space information which is quantized broadband radio wave information including the radio signals to the transceivers. Delivery mechanism consists of a distribution server which distributes radio space corresponding to the request of the client such as the center frequency and the bandwidth and a client which uses the radio space information. Accumulation of the distribution servers which deliver radio space information simultaneously to a large number of clients will contribute to build an infrastructure for any clients ubiquitously distributed over the globe. In this paper, scale-out architecture of a distribution server is proposed to deliver unlimitedly broadband radio space information to unlimited number of clients. Experimental implementation indicates the architecture to be a scale-out solution, while the number of clients is restricted by the computer resources of the distribution server. The band pass filter processing for individual client in the distribution server consumes the dominant part of the processing power, and the number of CPU cores is the upper limit of clients supportable for the distribution server in the current operating system implementation. The logical increase of the number of CPU cores by hardware multithreading does not contribute to relax this limit. We also discuss the guidance architecture or building server derived from these conclusions.