Databases Reference
In-Depth Information
Area
(
r, x
c
,y
c
), where a “random location” is a neighborhood of random point
(
x, y
)
Area
(
r, x
c
,y
c
). The probability that the real user's position (
x
u
,y
u
)
belongs to a neighborhood of a random point (
x, y
) is uniformly distributed
over the whole location measurement. Accordingly, the joint probability den-
sity function (pdf) of the real user's position can be defined as follows.
∈
Definition 3 (Uniform joint pdf).
Given a location measurement
Area
(
r, x
c
,y
c
)
, the joint probability density function (joint pdf) f
r
(
x, y
)
of
real user's position
(
x
u
,y
u
)
to be in the neighborhood of point
(
x, y
)
is:
f
r
(
x, y
)=
1
πr
if
(
x, y
)
∈ Area
(
r, x
c
,y
c
)
2
0
otherwise.
Before analyzing the obfuscation techniques in details, we first describe
how users can express their privacy preferences. Despite its importance for
the effectiveness of a privacy solution, this issue has received little attention in
previous works on location privacy. We then describe how the level of privacy
can be quantitatively expressed as a functional term independently from any
physical scale or specific technology.
4.1 User Preferences and Relevance Metric
Several works in location privacy field are based on the definition of users
privacy preferences by means of a minimum distance [7, 28]. This choice is
dictated by the fact that usually the users tend to adopt simple and intuitive
way for expressing their privacy preference and tend to be averse to complex
configurations. A user can define as her privacy preference a minimum dis-
tance, which results in a location area achieved by increasing the granularity
of the actual location measurement. In particular, assuming location measure-
ments as circular areas, the minimum distance privacy preference represents
the minimum radius of the area that a user is willing to release to other parties.
However, the definition of the minimum distance as user privacy preference
exhibits some shortcomings:
i)
it is highly dependent on the adopted privacy
solution;
ii)
it is suitable for only obfuscation techniques that increase the
granularity of the measurement;
iii)
it is dicult to integrate in a full-fledged
location-based application scenario [10, 33];
iv)
it is not suitable for solutions
using different obfuscation techniques.
To overcome these issues, others proposals [30, 31, 32] suggest a different
way to manage users privacy preferences. In these works, users specify their
privacy requirements through the definition of a
relative
degradation of the
location accuracy with respect to the location measurement, which is mod-
eled through an index
λ
∈
[0
,
∞
), where
λ
= 0 corresponds to no degradation,
λ
to maximum degradation, and intermediate values correspond to dif-
ferent degrees of degradation. For instance,
λ
=0.5 means 50% of degradation,
→∞
