Page 229 - ITU Kaleidoscope 2016

ICTs for a Sustainable World

(a) Overview of geo-location database approach (b) Overview of spectrum sensing

Figure 2: Overview of the TVWS identification approaches

in the area of interest is close to the measured data done in spectrum sensing, can be severe in the developed world be-
the same area; an Internet backbone infrastructure to facili- cause of many tall buildings, which are also close together.
tate efficient and frequent communication between a master Consequently, primary TV signals are difficult to detect ac-
WSD and a slave WSD and a master WSD and a geo-location curately by spectrum sensing alone and could result in harm-
database; existence of detailed centralised TV database infor- ful interference. Therefore, performance of spectrum sens-
mation in the area of interest. For the spectrum sensing ap- ing in urban areas may be low. On the contrary, rural ar-
proach to perform optimally, the following factors must ex- eas, especially those of developing world countries, where
ist: detection threshold value that is optimal such that there white space could be used to provide broadband connectivity,
is no harmful interference to the primary users or any missed are sparsely populated with small isolated traditional build-
opportunities by secondary users; minimal to no mult-path ing structures, which are unlikely to cause considerable fad-
fading and shadowing to avoid hidden user problem; large ing, shadowing or bring about the hidden user problem. In
blocks of TVWSs in the area of interest. terms of the Fresnel zone concept, these small isolated build-
ing structures could block the Fresnel zone to a maximum
obstruction allowable level of not more than 40%, which pro-
4. WHAT IS THE BEST APPROACH? duces little to no interference.

Since optimal performance of each approach is dependent Most rural areas of developing regions, for example sub-
on some factors that may not be present or exist in some re- Saharan Africa, have vast tracts of unused spectrum in the
gions, none can produce superior performance in all possible ultra-high frequency (UHF) band [9], which do not require
regions. For example, in regions or countries where prop- stringent restrictions like the developed regions. This makes
the use of the spectrum sensing approach a suitable alter-
agation models have been tried and tested extensively such
native because problems limiting its use in developed re-
that their prediction results are close to ground truth data,
gions may be considered to be much more forgiving than the
there is reliable Internet backbone infrastructure and a cen-
circumstances associated with dense urban areas. Further-
tralized detailed TV database information is also available,
more, cooperative spectrum sensing reduces errors in spec-
the geo-location database approach is expected to perform
trum sensing caused by multi-path fading and is a possible
better than spectrum sensing. That could be the case with
solution to the hidden user problem. Therefore, even these
developed regions such as the US and Europe where all the
problems may exist in the rural regions, the use of coopera-
three factors are present. It is therefore not surprising that the
tive spectrum sensing can improve its performance.
geo-location approach is being given preference as the main
technique of finding TV white spaces [8] in those regions. There are some additional factors that can come into play
However, conditions in developing regions are quite different when deciding which technique to use, apart from the above-
from developed regions. Internet backbone infrastructure is mentioned performance factors. One such metric is the cost
poor and unreliable, more especially in the rural areas; prop- to implement, maintain, and administer the approaches. Geo-
agation models have rarely been tested here such that their location approach requires a complex centralized structure
behaviour is unclear at the moment; spectrum usage infor- and even more complex logistics, and as such, its imple-
mation is scattered and stored in many formats, electronic mentation cost, maintenance cost and administrative cost is
and paper, and the regulators have not collected it into a use- higher than spectrum sensing approach [8]. This too is likely
ful centralized database that is publicly available. Therefore, to make spectrum sensing a more favourable approach to the
at these prevailing conditions, the use of a geo-location spec- developing world as most countries in this region may not
trum database approach may not produce optimal results. have the necessary infrastructures present and must be built
Fading, shadowing and the hidden user problem, relevant to especially for the DSA, which is expensive.

– 211 –