Wireless Networking in Higher Education Vol.

2, 2002

6
Installing Wireless Networks

Wireless network installation entails more with specific colleges and departments to
than placing random access points through- address their individual research and aca-
out a building; the network is actually an demic requirements. (See Table 6-1.) This is
intricate web of transmission points specifi- especially true as an institution’s full-time
cally placed to ensure the most effective equivalence (FTE) grows. Dan Shapiro, direc-
transmission pattern. As a result, network tor of information services and chief infor-
design relies almost as much on intuition as mation officer, School of Dental Medicine,
on proven design techniques. University of Pennsylvania, explained, “The
school wanted better network access for
Planning/Implementation students’ research and clinical activities.
Participants The university’s central IT department is
IT departments initiate many wireless working with us to help bring this about.”
networks, but frequently they work in tandem Smaller institutions may rely on the IT

Table 6-1. Academic Groups Initiating Wireless Network, by FTE Size*

FTE: 1– FTE: 5,000– FTE: 10,000–

Total 4,999 9,999 19,999 FTE: 20,000+
Users (N=329) (N=170) (N=53) (N=52) (N=21)

IT Department 89% 87% 89% 96% 86%

Specific College/Departmental 24% 15% 23% 44% 52%

Library 20% 14% 23% 35% 24%

Faculty 19% 19% 17% 25% 19%

Administration 15% 14% 21% 10% 24%

Students 6% 5% 4% 6% 0%

Research Centers 4% 2% 4% 8% 10%

* As reported by current/planned wireless network operators. Academic departments most often mentioned by
respondents are highlighted.

© 2002 EDUCAUSE. Reproduction by permission only.

EDUCAUSE CENTER FOR A PPLIED R ESEARCH 53

Wireless Networking in Higher Education Vol. 2, 2002

Table 6-2. Academic Groups Involved in Implementation Process, by Carnegie

Classification*

Total Doctoral Master’s Bachelor’s Associate’s

Group Involved (N=329) (N=67) (N=80) (N=71) (N=32)

IT 98% 99% 95% 100% 100%

Administration 43% 34% 49% 35% 59%

Library 53% 58% 53% 61% 41%

Research Center 10% 25% 6% 3% 0%

Specific College or Dept. 43% 67% 46% 24% 41%

Faculty 40% 46% 38% 31% 47%

Students 19% 22% 18% 21% 6%

IT Manufacturers 18% 19% 16% 13% 19%

Communications Provider 17% 15% 21% 17% 9%

System Vendor 25% 25% 31% 27% 19%

* As reported by current/planned wireless network operators. The non-IT academic department most often
mentioned by respondents is highlighted.

department as a more centralized resource. IT departments are almost always highly in-
Larry Levine, director of computing, volved in planning and implementing wire-
Dartmouth College, stated, “The wireless less networks. Again, specific departments
idea came from the department/faculty level. and colleges are more likely to be involved
The central IT organization picked up the at doctorate institutions, reflecting their
ball, making the necessary arrangements. larger scope. David Ferriero, vice provost and
The user believes that the IT organization university librarian, Duke University, said,
is eager to hear of ideas that they can sup- “The planning is coordinated between the
port. The faculty sees this as the major ben- enterprise IT group and various groups and
efit.” Richard Parker, CIO and director of departments on campus. Departments are
computing and information services, Harvey consulted to determine coverage needs.”
Mudd College, concurred: “IT plans, main-
tains, and updates the wireless network. If Planning/Implementation
an academic department would like to ex- Characteristics
pand or enhance coverage for some of their Table 6-3 summarizes the wireless net-
space, we discuss their needs with them work planning and implementation charac-
and move in recommended directions as teristics of online survey respondents at
the budget permits.” different institutions by Carnegie classifica-
The IT department continues its involve- tion. The complex nature of many doctoral
ment throughout the implementation pro- institutions has an interesting impact on the
cess, partly to ensure interoperability process: implementation time is longer, costs
around the campus. As Table 6-2 illustrates, are greater, and more students are served.

54
Wireless Networking in Higher Education Vol. 2, 2002

Table 6-3. Planning and Implementation Characteristics of Respondents, by

Carnegie Classification*

Total Doctoral Master’s Bachelor’s Associate’s

(N=299) (N=64) (N=75) (N=62) (N=28)

Median Planning Time (Months) 6 6 6 6 6

Median Implementation
Time (Months) 3 5 2 2 1

Median Amount Spent

or Budgeted $50,000 $80,000 $50,000 $20,000 $30,000

Median Number of Students/

Faculty That Wireless
Network Serves 1,000 3,000 1,000 610 500

Median Cost/User $50.00 $26.67 $50.00 $32.79 $60.00

* Base: Current/piloting wireless network operators.

Yet their large scope enables doctoral insti- less network implementation. Sometimes IT
tutions to achieve economies of scale. Their departments will fund the original design
median cost per user is far lower than that and installation of the wireless network to
of other institution types. seed its adoption. Making the resource avail-
able for free encourages academic colleges,
Source of Funding schools, and departments to experiment
As Table 6-4 illustrates, the IT depart- with wireless technology with little financial
ments of approximately 60 percent of online investment. “They will come, but we have
survey respondent institutions fund the wire- to make it reasonable pricewise for users to

Table 6-4. Wireless Network Implementation Funding Source, by Carnegie

Classification*

Source of Total Doctoral Master’s Bachelor’s Associate’s

Implementation Funding (N=329) (N=67) (N=80) (N=71) (N=32)

Overall IT Budget 57.0% 45.2% 55.5% 67.7% 52.5%

Specific Department 11.9% 19.3% 10.0% 4.6% 9.7%

Specific Academic or
Research Grant 9.3% 4.8% 9.3% 12.3% 13.6%

IT Supplier Donation 1.9% 3.3% 0.7% 0.5% 0.0%

Student Usage Fees 7.5% 11.8% 10.7% 3.8% 7.3%

Other 11.4% 15.6% 13.3% 9.5% 14.8%

* Mean percentage as reported by current/planned wireless network operators. The non-IT source most often
mentioned by respondents is highlighted.

EDUCAUSE CENTER FOR A PPLIED R ESEARCH 55

Wireless Networking in Higher Education Vol. 2, 2002

get in. So we subsidized wireless in the be- Installation Issues

ginning by funding the infrastructure and by Installing a wireless network is not as easy
providing the loaner equipment in the library,” as one, two, three. “We put a fair amount
explained Ken Blackney, core technology in- of effort up front in the designs to create as
frastructure director, Drexel University. stable a network as possible to avoid paying
But as wireless becomes operational, the on the back end with the support,” stated
IT department continues to fund its opera- Chuck Bartel, director of operations and
tions at many institutions. (See Table 6-5.) project director of Wireless Andrew,
In the long term, this may raise issues, as Carnegie Mellon University. “Institutions
institutions struggle to create a cost-recov- may try a quick and easy design by throw-
ery model. “We had a difficult time putting ing some access points here or there, but
our hands around it, because how do you they’ll spend significant time after the fact
bill for it? How can you manage who’s us- troubleshooting problems.”
ing it?” explained Judy Huddleston, direc- Many online survey respondents planned
tor, IT infrastructures administrative services, extensively; over half (57 percent) conducted
The University of Tennessee. Joel Smith, chief a radio frequency (RF) site study. Doctoral
information officer, Carnegie Mellon Univer- and master’s institutions are much more
sity, advised institutions to “think through likely to have done so, perhaps indicating
the costs before you start, because you are the more complex nature of their network-
really building a parallel network. Your wired ing environment and/or greater IT resources.
network is not disappearing, so you are The split is fairly even between those using
making a significant add-on. You’ve built an internal resources (31 percent) and those
entire new infrastructure that requires re- using external resources (26 percent) to
freshing, and it has a technological improve- conduct the study.
ment cycle. That means you really have to Of those doing the study internally,
build a cost-recovery model.” roughly one-third possessed the expertise

Table 6-5. Wireless Network Maintenance and Operational Funding Source, by

Carnegie Classification*

Source of Total Doctoral Master’s Bachelor’s Associate’s

Implementation Funding (N=329) (N=67) (N=80) (N=71) (N=32)

Overall IT Budget 86.0% 81.0% 81.0% 99.0% 84.0%

Specific Department 7.0% 15.0% 5.0% 1.0% 3.0%

Specific Academic or
Research Grant 16.0% 39.0% 19.0% 3.0% 6.0%

IT Supplier Donation 6.0% 13.0% 4.0% 3.0% 13.0%

Student Usage Fees 4.0% 6.0% 4.0% 3.0% 0.0%

Department/School Usage Metric 3.0% 3.0% 4.0% 1.0% 6.0%

Other 11.0% 13.0% 20.0% 8.0% 3.0%

* Base: Current/planned wireless network operators. The non-IT source most often mentioned by respondents is
highlighted.

56
Wireless Networking in Higher Education Vol. 2, 2002

already. (See Figure 6-1.) Most (93 percent) maximize coverage and capacity. Doctoral
of those using an external contractor to con- and 20,000-plus FTE institutions (more than
duct the site study said the contractor had a 40 percent in both cases) were much more
good track record. Almost all (95 percent) likely to do so, again perhaps indicating the
were satisfied with the external contractor’s more complex nature and greater scope of
service. their wireless networks. Those that imple-
Institutions used different methods for se- mented an antenna scheme used a variety
lecting their wireless equipment vendors. of methods. Individual responses follow:
Some preferred to use vendors from previ- ◆ Used varied antenna types and radio
ous applications. Greg Schaffer, director of power, including Aeronet 350s, which
network services, Middle Tennessee State have two coax antenna connections. Un-
University, explained, “We went with a ven- der continual adjustment.
dor who handles our wired network compo- ◆ Chose various schemes, depending on the
nents, for compatibility, familiarity with coverage requirements and expected den-
company and personnel, and functionality.” sity of wireless users.
Others compared vendors. Barry Walsh, ◆ Involved internal RF experts, but largely
director, Indiana University, stated, “We tested determined the scheme and placement
three vendors’ access points and PC cards and empirically.
determined that interoperability was as prom- ◆ Selected antennae to match the size and
ised. We picked one vendor based on minor shape of the area. For several outside loca-
preferences for its management application.” tions, used directional high-gain antennae.
Dewitt Latimer, executive director, IT infra- ◆ Used a combination of technologies,
structure, The University of Tennessee, said, based on the situation, most notably de-
“We created a list of [five] criteria for vendor ployment of radiating coax cable.
selection: security, upgradability to [IEEE] ◆ Installed low-gain (2 dbm) dual-diver-
802.11a, 54-megabit-per-second spec, cost sity antennae for small, isolated areas,
of ownership over three years, and willing- medium-gain (6 dbm) omnidirectional
ness of the company to work with UT as far antennae with overlapping cells to pro-
as a partnership.” vide wide-area coverage inside build-
Fewer than one-third of the respondents ings, and high-gain omnidirectional
installed a specific antenna scheme to (8 dbm) antennae for outside coverage.

* Base: Current/piloting wireless network operators conducting RF study internally (N = 92).

Figure 6-1. Expertise for Conducting RF Site Study*

EDUCAUSE CENTER FOR A PPLIED R ESEARCH 57

Wireless Networking in Higher Education Vol. 2, 2002

◆ Selected various antennae as required to a large number of simultaneous users, but

improve coverage, mostly by trial and if you put an access point in every class with-
error. out the proper antenna, you’ll have signal
◆ Used a variety of antenna configurations interference.”
to give maximum coverage. AC circuits power the access points for
◆ Deployed access points with standard an- 62 percent of online respondents; the other
tenna configurations to deliver overlap- 38 percent use power over the Ethernet
ping coverage areas, added additional cable. Older wireless network installations
access points in locations with larger may rely on AC power because power via
numbers of users. the Ethernet is a relatively recent design
◆ Chose overlapping zones with omnidi- option. And when refreshing access points,
rectional antennae tuned to specific institutions may switch to power over
fields. Ethernet because it eliminates one power
Two institutions provided examples of the outlet drop, saving considerable installation
multidimensional aspects of wireless net- expense. Before he decided to use power
work installation. over the Ethernet directly, Ken Blackney,
Richard Parker, Harvey Mudd College, Drexel University, was “getting estimates of
described his installation process: “Harvey $1,100 per outlet for electricity.”
Mudd has relatively new buildings (for ex-
ample, no 4-foot-thick walls, few solid inte- Wireless Standards
rior concrete walls) that are not spaced Table 6-6 illustrates online survey respon-
particularly close together. This simplifies dents’ current and future technology stan-
things. In general, given the wireless stan- dards plans. IEEE 802.11b/Wi-Fi is the
dard, we planned coverage for each build- dominant technology today, but fewer re-
ing, tested it for the initial targeted areas, spondents plan to continue supporting it 24
and made adjustments in the antenna place- months from now, as more institutions
ment (or, occasionally, [antenna] type) as switch to 802.11a. Support for 802.11b/Wi-
necessary. By planning for the whole build- Fi varies by Carnegie classification because
ing, [issues of] range, coverage, and spec- the subsequent equipment investment may
trum allocation were relatively deter smaller institutions from switching.
straightforward to address. Coverage for “802.11b equipment is getting really inex-
nearby outdoor spaces was planned in con- pensive. There are still many situations where
junction with the building.” the convenience of wireless connectivity
Mark Cain, executive director, informa- makes the lower bandwidth of 11.b not an
tion services and support, Mt. St. Joseph issue. We will continue to deploy it as the
College, explained, “We followed vendor application makes sense,” said KC Hundere,
recommendations. You have 11 channels director of network services, Glendale Com-
available, but effectively can use only three, munity College.
because you have to have three- to four- Other institutions are debating the pros
channel separation between nearby access and cons of an 802.11a upgrade. Deke
points. Then you have to think three-dimen- Kassabian, senior technology director, Uni-
sionally. Range, coverage, and spectrum use versity of Pennsylvania, is concerned with
are probably best managed with the right faster transmission and an uncrowded spec-
kind of antenna. In a classroom, you need trum. “We are interested in 802.11a, partly
to be able to provide a lot of bandwidth for in order to gain higher throughput,” he

58
Wireless Networking in Higher Education Vol. 2, 2002

Table 6-6. Current and Projected Support for Various Wireless Standards, by
Carnegie Classification*

Total Doctoral Master’s Bachelor’s Associate

(N=299) (N=64) (N=75) (N=62) (N=28)

Add in 24 Add in 24 Add in 24 Add in 24 Add in 24

Now Months Now Months Now Months Now Months Now Months

802.11b/Wi-Fi 90% 67% 91% 56% 89% 65% 90% 76% 93% 82%

802.11a 13% 53% 13% 70% 11% 56% 15% 42% 18% 32%

802.11g 4% 25% 2% 31% 3% 24% 6% 19% 7% 14%

Bluetooth 3% 18% 2% 22% 3% 16% 2% 15% 7% 18%

Broadband 1% 6% 0% 5% 0% 5% 0% 5% 0% 11%

* Base: Current/piloting wireless network operators.

explained, “but mostly to get out of McCredie’s concerns are justified be-
the crowded 2.4-GHz spectrum.” This is a cause the IEEE standards committee is
special concern to avoid interference with about to approve the 802.11g standard,
Bluetooth-enabled devices. adding to the standards confusion. This
Lack of 802.11a/802.11b compatibility standard addresses incompatibility issues
is another issue. Greg Schaffer of Middle by functioning like a step-down modem,
Tennessee State University explained, transmitting at both the 2.4-GHz and 5-
“We’ve deployed access points that can GHz levels. Chuck Bartel of Carnegie
handle both 802.11b and 802.11a clients Mellon University believes that 802.11g
simultaneously for future needs, but [we] products will not reach the market until
only offer 802.11b connectivity now. With late 2002 at the earliest. When they do,
the dual capability, we’ll be able to offer he predicts that “many institutions will op-
802.11a service without alienating our erate bimodal networks, supporting both
802.11b customer base. We’re also consid- 802.11a and 802.11b. When 802.11g ar-
ering using 802.11a to connect remote rives in the marketplace, they will replace
access points that cannot be connected to their 802.11b cards with 802.11g cards be-
the wired network directly.” cause the 802.11g card can step down to
Standards turbulence is a concern for 802.11b mode.”
Jack McCredie, CIO and associate vice chan- Most IT administrators believe the wire-
cellor, information systems and technology, less network has minimal impact on net-
University of California, Berkeley. He said, work traffic in general. Few cite any impact
“We will deploy 802.11a when we feel it at all. Mark Cain of Mt. St. Joseph Col-
has matured enough to be reliable and the lege explained, “There is an increase in
costs are such that we can reasonably re- traffic, but the wired network is far faster
place the equipment on a standard three- than the wireless, so it doesn’t even
or four-year life cycle.” breathe hard.”

EDUCAUSE CENTER FOR A PPLIED R ESEARCH 59