Workshop UWBAN

Research Article

Generic Small Scale Channel Model for On-Body UWB WBAN Communications

Download577 downloads
  • @INPROCEEDINGS{10.4108/icst.bodynets.2013.253673,
        author={Timo Kumpuniemi and Matti H\aa{}m\aa{}l\aa{}inen and Tommi Tuovinen and Kamya Yekeh Yazdandoost and Jari Iinatti},
        title={Generic Small Scale Channel Model for On-Body UWB WBAN Communications},
        proceedings={Workshop UWBAN},
        publisher={ICST},
        proceedings_a={UWBAN},
        year={2013},
        month={10},
        keywords={body area network small scale channel model ultra wideband},
        doi={10.4108/icst.bodynets.2013.253673}
    }
    
  • Timo Kumpuniemi
    Matti Hämäläinen
    Tommi Tuovinen
    Kamya Yekeh Yazdandoost
    Jari Iinatti
    Year: 2013
    Generic Small Scale Channel Model for On-Body UWB WBAN Communications
    UWBAN
    ICST
    DOI: 10.4108/icst.bodynets.2013.253673
Timo Kumpuniemi1,*, Matti Hämäläinen1, Tommi Tuovinen1, Kamya Yekeh Yazdandoost1, Jari Iinatti1
  • 1: Centre for Wireless Communications, University of Oulu
*Contact email: timo.kumpuniemi@ee.oulu.fi

Abstract

This paper presents a generic small scale channel model for ultra wideband wireless body area network communications. It is based on static on-body measurements in an anechoic chamber by using a vector network analyzer within a 2-8 GHz bandwidth. Two antenna types are used for the examination: dipole and double loop. From the existing data, a generic average channel impulse response (CIR) was extracted resulting 11 and 12 resolvable paths for the dipole and the double loop antenna, respectively, when a dynamic range of 25 dB was used. The CIR envelopes were modelled by using the polynomial least squares (LS) fitting with orders one to five. The 5th order LS model was noted to follow the CIR envelope most precisely. The CIR decays slightly faster for the dipole antenna. The statistical properties of the CIR bins were solved by fitting the data for 17 continuous distributions and ranking them by using the second order Akaike information criterion. To model the CIR amplitudes exactly, four different distributions were needed for the dipole but for an approximate model Weibull and lognormal distributions suffice. For the double loop, all CIR bins follow the inverse Gaussian distribution. The distributions of the CIR bin indexes were found to follow the negative binomial distribution for both antennas.