Wireless Local Area Network Positioning - Ambient Intelligence for Scientific Discovery - page 255

Fig. 3. User centric map of Wireless Andrew shows the locations of all mobile users
connected to the network.

on campus, color-coded to represent the number of users connected to it, and a
map that displays the location of every access point on campus. Figure 3 shows
a screen shot of the user-centric map. A quick glance at the map shows the
dispersion of students throughout the campus. Clicking on the Graduate School
of Industrial Administration building tells us that there are 337 users connected
to 18 access points with an average signal to noise ratio (SNR) of 18.72 (Fig. 4).
Further, and significantly more detailed, comparative and historical statistical
analysis on the network and its clients was performed as well and are featured
on the cmuSKY web site.

4.1 Optimizing a Wireless LAN

The substitution of tangible cable with invisible radio wave has made the state
and nature of wireless networks somewhat nebulous. Whereas the physicality of
a wired network defines its bounds, wireless networks have indistinct boundaries
which change from user to user, creating an array of logistic and security issues.
Designing a wireless local area network is a highly empirical process requiring
substantial testing and repositioning of access points and there are no rule-of-
thumb heuristics that necessarily constitute good placement of wireless access
points. This largely results from the fact that administrators do not know, up-
front, the concentrations of wireless devices to expect in different areas of the
network. As a result, administrators must carefully plan for both bandwidth
capacity and RF coverage in what is generally an ad-hoc approach. To add to
the challenge, administrators must account for the fact that they cannot contain
the reach of their networks, thereby enhancing the possibility of unauthorized