Databases Reference
In-Depth Information
λ
=1 means 100% of degradation and any value
λ>
1 corresponds to a degra-
dation greater than 100%. Although both minimum distance
d
and index
λ
are easy to specify for users,
λ
is a more general solution because independent
from a specific location measurement and obfuscation technique. However,
the definition of
λ
is not sucient, especially when we need to balance the
users needs of privacy and the LBSs needs of location accuracy to maintain
an acceptable quality of the online service.
To accommodate the peculiar characteristics of the above scenario, the
concept of
relevance
is introduced as the adimensional metric of both the
accuracy and the privacy of a location information, abstracting from any
physical attribute of sensing technology. A relevance
is a value in (0,1]
associated with each location information, which depends on measurement
errors and privacy preferences of users. In particular,
R
tends to 0 when the
location information is considered unreliable for service provision;
R
=1 when
the location information is equal to the original location measurement;
R
R∈
(0,1) when the location information has various degrees of accurateness. The
location privacy associated with an obfuscated location is evaluated by (1-
).
Applying the concept of relevance to a LBAC scenario, an LBAC service
has to manage the following different relevances:
R
•
Technological relevance
(
R
Tech
) is the metric for the accuracy of the loca-
tion measurement provided by a location service given a mobile technology
and its technical quality.
•
Privacy relevance
(
R
Priv
) is the metric for the accuracy of an obfuscated
location and therefore the level of privacy provided to the users.
•
LBAC relevance
(
R
LBAC
) is the metric for the lowest accuracy of the
location information that an LBAC service is willing to accept. It is re-
quired by the business application for a location measurement or for a
location-based predicate evaluation.
•
Evaluation relevance
(
R
Eval
) is the metric for the accuracy of a LBAC
predicate evaluation.
R
Tech
are assumed to be known.
R
Priv
is derived from the privacy preferences expressed by users, while
Among these relevances,
R
LBAC
and
R
Eval
is calculated by the system (see Sect. 5). In other words,
R
Priv
represents
the relevance of the final obfuscated area that is calculated starting from the
location measurement with relevance
R
Tech
and by degrading its accuracy
according to the value of
λ
. Formally,
R
Priv
is calculated as:
R
Priv
=(
λ
+1)
−
1
R
Tech
(1)
If a privacy preference is expressed through a minimum distance
r
,itis
straightforward to derive
λ
from
r
. The obfuscated area is then calculated
by scaling up the radius of the location measurement until the user privacy
preference
λ
is satisfied.
