Wireless Local Area Network Positioning - Ambient Intelligence for Scientific Discovery - page 258

Fig. 6. GSIA Students actively using laptops.

5.3 Favorite Sites

In addition to a home-node, many users have a collection of favorite sites, defined
as locations where they exhibit a high frequency of usage. Often, users follow set
routines that take them to a handful of places throughout the day. For exam-
ple, a student's class schedule can provide a predictable pattern of movement.
Similarly, school faculty have a set schedule in which time is split between labs,
class and an oce.
To determine and rank the favorite sites, the time spent by each MAC address
at each access point was calculated. We then took the top five access points,
ranked by time, for each user; this includes the home-node discussed earlier. The
results show an overall lack of mobility. Only 3% of users spent less than 50%
of their time at their top 5 sites. This contrasts with the fact that 73% of users
spent 90% or more of their time at their top 5 favorite locations. Consequently,
the majority of users are mainly concerned about network performance in a
handful of locations.
As with the home-node classification, the two user communities also exhibit
different usage patterns. For GSIA users, the median percentage of time spent
at the top five sites is 92.1%; for non-GSIA users it is 99.1%. This difference is
not as great as exhibited for the home-node differential, however, the important
feature to note is how little time all users spent away from their top five sites.
For the overall user community, slightly over one-third of users spent no time
outside of their top five favorite sites.

5.4 Mobility

The absolute metric of mobility is defined as the number of access points a
user connects to over a given time period. Stationary users will exhibit mobility
values of 1 while high movement users will have values hovering around 10+.
These mobility values are calculated by counting the number of unique access
points that each unique MAC address accessed for each day.
The mobility results in Fig. 7 show that the majority of users are not highly
mobile. Only 10.5% of all users have average mobility values of 5 or above. The
median mobility for all users is 2.1.