9th International Conference on Cognitive Radio Oriented Wireless Networks

Research Article

When Primary Users Whisper: A Preliminary Analysis on Correlations of Population-Traffic Dynamics

Download484 downloads
  • @INPROCEEDINGS{10.4108/icst.crowncom.2014.255403,
        author={Alexandros Palaios and Maria Michalopoulou and Janne Riihij\aa{}rvi and Petri M\aa{}h\o{}nen},
        title={When Primary Users Whisper: A Preliminary Analysis on Correlations of Population-Traffic Dynamics},
        proceedings={9th International Conference on Cognitive Radio Oriented Wireless Networks},
        publisher={IEEE},
        proceedings_a={CROWNCOM},
        year={2014},
        month={7},
        keywords={population-traffic correlations spectrum measurements},
        doi={10.4108/icst.crowncom.2014.255403}
    }
    
  • Alexandros Palaios
    Maria Michalopoulou
    Janne Riihijärvi
    Petri Mähönen
    Year: 2014
    When Primary Users Whisper: A Preliminary Analysis on Correlations of Population-Traffic Dynamics
    CROWNCOM
    IEEE
    DOI: 10.4108/icst.crowncom.2014.255403
Alexandros Palaios1, Maria Michalopoulou1, Janne Riihijärvi1,*, Petri Mähönen1
  • 1: RWTH Aachen University
*Contact email: jar@inets.rwth-aachen.de

Abstract

In this paper we report the preliminary results from the large-scale spectrum measurement campaign based study to find out correlations between spectrum use and socio-economical data. We use two large spectrum usage datasets that have been collected from metropolitan London and Paris, and correlate the spectrum usage of GSM, UMTS, and 2.4 GHz ISM band against the local population density and location type in those cities. We show that although the spectrum usage, measured in duty cycle and average received power in bands, in general shows the correlation with the population density, one has to also consider the type of the city area as an extra parameter. One of the key surprising findings is that that there are also areas, where the spectrum usage is uncorrelated, or even negatively correlated, with the increasing population density. This shows that spectrum usage models need to be more detailed than just population density conditioned. Moreover, the measurement campaign and found correlations show that one can derive useful information on the network structure and conduct useful test driving campaigns with a surprisingly small amount of data points.