Page 233 - ITU Kaleidoscope 2016

ICTs for a Sustainable World

Table 6: RMSE between measured and models’ estimation and administrative cost of geo-location approach, we have
power concluded that spectrum sensing is more favourable to use in
rural areas of developing world countries than geo-location
Propagation Model RMSE (dB)
FSPL 6.06 database approach.
L-R 9.64 Since cooperative spectrum sensing eliminates the limitation
Ericsson 9999 24.15
Egli 28.64 of local spectrum sensing, the authors would like to
Hata 45.80 investigate a white space network detection framework
based on cooperative spectrum sensing using hand-held
spectrum analyzers as part of their future work, and also
5.6. Discussion of the results
extend the analysis to find the available white spaces at
UWC measurement site, which could be used for campus
From the plots of the path losses in Figure 4 and the path
networking.
loss errors in Table 5, the FSPL is the closest model to the
measurements. The accurancy of the FSPL can be attributed
the possiblity of clear line of site between the TV transmitter REFERENCES
and the measurement locations since the measurement loca-
tions were just few kilometers away from the TV transmitter
[1] I.F. Akyildiz, W. Lee, M.C. Vuran, and S. Mohanty,
site. The L-R (ITM), using Radio Mobile [13], was also ac-
“Next generation/dynamic spectrum access/cognitive
curate as it uses terrain elevation data of an area to compute
radio wireless networks: a survey,” Computer net-
the path loss. The Ericsson 9999, Egli and Hata underes-
works, vol. 50, no. 13, pp. 2127–2159, 2006.
timated the received signal power such that their estimated
path losses are greater than the derived path losses from the [2] R. Kaniezhil, C. Chandrasekar, and S. NithyaRekha,
measurements at each location. “Performance evaluation of qos parameters in dynamic
Although the FSPL model is closest to the measurement data spectrum sharing for heterogeneous wireless communi-
as its RMSE (6.06 dB) from Table 6 falls within the accept- cation networks,” 2012.
able range of 6-7 dB for urban areas [14], we cannot con-
clusively say that is the best-fit model for the area within [3] Y. Yuan, P. Bahl, R. Chandra, T. Moscibroda, and
the distances where the measurements were taken. Exten- Y. Wu, “Allocating dynamic time-spectrum blocks in
sive long-hours spectral measurements are needed and also cognitive radio networks,” in Proceedings of the 8th
more measurement sites need to be included to confirm the ACM international symposium on Mobile ad hoc net-
validity of a model for the area, which may be costly and working and computing. ACM, 2007, pp. 130–139.
time consuming.
[4] J. Carlson, N. Ntlatlapa, J. King, F. Mgwili-Sibanda,
H. Hart, C. Geerdts, and S. Song, “Studies on
6. CONCLUSION AND FUTURE WORK the use of television white spaces in south africa:
Recommendations and learnings from the cape
Discussions in the literature have placed much emphasis on town television white space trial,” electronic,
the limitations of the spectrum sensing approach and mainly https://www.tenet.ac.za/tvws/recommendation-and-
based on the developed world environment ignoring the lim- learnings-from-the-cape- town-tv-white-spaces-trial,
itations of the geo-location database approach and its impact 2013.
on the developing world. This paper considered a broader
[5] T. Y¨ ucek and H. Arslan, “A survey of spectrum sensing
analysis of the approaches by looking at factors that can af-
algorithms for cognitive radio applications,” Commu-
fect the performance of each approach and how the presence
nications Surveys & Tutorials, IEEE, vol. 11, no. 1, pp.
or absence of these factors in a developed region and/or in
116–130, 2009.
a developing region can affect their performance. In so do-
ing, the paper has highlighted the need to conduct more re-
[6] T. Brown, E. Pietrosemoli, M. Zennaro, A. Bagula,
search on the performance of spectrum sensing in regions
H. Mauwa, and S. Nleya, “A survey of TV white space
where plenty of white spaces are available, e.g., rural areas
measurements,” in e-Infrastructure and e-Services for
of developing world countries.
Developing Countries. 2014, pp. 164–172, Springer In-
Our analysis shows that information that is needed by the
ternational Publishing.
geo-location database approach to perform optimally may
not exist in most developing world countries especially in the [7] L. Doyle, Essentials of cognitive radio, Cambridge
rural areas where telecommunication infrastructure is lack- University Press, 2009.
ing and white spaces are abundant. Based on the absence
of factors that are needed by the geo-location approach to [8] V. Gonc¸alves and S. Pollin, “The value of sensing for
perform optimally in developing world regions, presence of TV white spaces,” in New Frontiers in Dynamic Spec-
abundant white spaces in rural areas of developing world re- trum Access Networks (DySPAN), 2011 IEEE Sympo-
gions and also considering the implementation, maintenance sium on. IEEE, 2011, pp. 231–241.

– 215 –