Database Reference
In-Depth Information
voyage103 01:A3:B5:0A:4B:42 rssi.log
20100720-175338-CEST,20:21:A5:45:40:40,5898756,-81
20100720-175340-CEST,20:21:A5:45:40:40,5898756,-80
20100720-175341-CEST,20:21:A5:45:40:40,5898756,-72
20100720-175353-CEST,20:21:A5:45:40:40,5898756,-78
20100720-175355-CEST,20:21:A5:45:40:40,5898756,-82
↓
voyage103 01:A3:B5:0A:4B:42 scan.log
20100720-175338-CEST,20:21:A5:45:40:40,5898756,in
20100720-175341-CEST,20:21:A5:45:40:40,5898756,out
20100720-175353-CEST,20:21:A5:45:40:40,5898756,in
20100720-175355-CEST,20:21:A5:45:40:40,5898756,out
Figure 14.2 Extract of logged data showing the raw time-point detection data (top) and the
compressed time-interval data (bottom), depicting the compression of solitary detections
into intervals leading to an abstract and structured geo-localized trajectory. This example
shows one Bluetooth device (MAC address 20:21:A5:45:40:40) being detected five times.
The buffer time of 10 seconds causes the raw data to be split into two separate detection
time intervals (in
→
out). The COD code of the device (5898756) shows that this was a
cell phone.
of log files during the scanning process in the following compressed format:
timestamp of detection, MAC address of the detected device, COD code of
the detected device, in/out/pass
. A buffer time of 10 seconds is used to create
detection time intervals from the detection time points.
In
is written when
a device enters the detection range of the sensor, and
out
is written when the
device leaves the range.
Pass
is used for solitary detections with no prior or later
detections within 10 seconds. The principle of this logging system is depicted in
Figure
14.2
. In correct terminology, this compression actually transforms a geo-
localized semantic trajectory into an abstract and structured semantic trajectory
where individual detections are compressed into detection intervals representing
the presence of a mobile device within a scanner's range during a certain time
interval.
This compressed interval-based representation adhering to the proximity prin-
ciple is then imported into our processing environment for further analysis.
Figure
14.3
shows a screenshot of this environment, dubbed a Geographical
Information System for Moving Objects (GisMo). It was developed in Java as a
desktop client.
14.3 Case Studies
To give a general overview of the merits of the Bluetooth tracking methodology,
we will show three case studies that have been carried out in three different
application contexts: crowd management and safety at a mass event, and mar-
keting insights in two retail environments: a professional fair and a shopping
mall.
