How does an institution determine whether a campus-wide wireless deployment is viable? Drawing from her experiences in planning Carnegie Mellon’s wireless LAN, Tracy Futhey offers factors and considerations that may be helpful in deciding.

Carnegie Mellon has deployed a campus-wide Wireless LAN (WLAN), which is serving 2,000 students, faculty and staff. The network has enabled the use of technology in places where it was not previously envisioned, and has been viewed as a highly successful project by faculty, students, and staff alike. But every university is different, and what has worked at Carnegie Mellon may not be desirable or feasible for another institution.

Evaluate Drawbacks and Benefits

For just about every positive aspect of wireless, there can be a corresponding downside. We'll begin with those factors.

From the computing operations standpoint, deploying a wireless network means managing and supporting two networks (since it is highly unlikely that a campus would choose to deploy only wireless). As a shared medium, the bandwidth available to a given wireless user will degrade with an increasing number of users who are working in close proximity, or when one or more users is performing particularly network-intensive tasks (such as streaming media).

Beyond the operational considerations are potential issues related to security and "ownership" of the radio frequencies used for wireless as well as other technologies. While security is important to all of us, those campuses located in densely populated urban areas may have need for higher levels of security than the current generation of wireless products provide. For example, NYU has been eager to deploy a WLAN for two years, but the lack of sufficient security has been a major impediment. Existing and emerging technologies such as VPNs (providing encryption using Virtual Private Networks), 802.1x (an IEEE specification for port-based access control) and PPPoE (an IETF specification for establishing point-to-point connections over Ethernet) show promise for increasing security over wireless.

There are also issues related to radio frequency interference from both off- and on-campus sources. Since no one actually owns the airspace in which WLANs operate, any number of devices can cause interference. For example, at MIT there is so much faculty research and other work related to wireless that the Information Systems group encourages faculty to use just one of the three available wireless channels, and then they will attempt to design around that channel as much as possible.

One final potential drawback is the extent to which wireless can be a distraction in the classroom. We have used wireless laptops and handheld devices at Carnegie Mellon in dozens of classes. While students and faculty are positive about the experience overall, if asked to identify negative aspects, the most common response is that a networked computing device in the classroom can be distracting.

Turning to the positive side, wireless networks can introduce a great deal of flexibility to the classroom, assuming that sufficient numbers of students have laptops. And if students often work in groups on projects and assignments, wireless-enabled laptops or handheld devices are a natural extension to those group exchanges, and save time in documenting meeting notes or collaborating on lab reports. In some cases, wireless can be used in place of wired network connections. For example, a small classroom can often be "connected" as effectively by using wireless as with a traditional wired network outlet for every student seat, and at a much lower cost.

The ability to be connected anywhere and any time has two other important potential benefits-increased productivity and increased satisfaction for users. Because individuals can be connected at their convenience, there are more opportunities to be productive (such as the time spent waiting to meet a friend). Also, since people can be connected at the location they choose, there is an increased sense of freedom and control over their work.

Identify the Costs

As with any potential deployment of new technology, one needs to consider the costs along with the anticipated benefits. These costs can vary greatly depending on the size and layout of the campus, the scope of the deployment, and the construction of the campus buildings. Startup costs for a WLAN include design, equipment purchase, and installation; ongoing costs include maintenance of the wireless network and user support.

Design costs may be direct or indirect, depending on whether your institution chooses to outsource or design in-house. The actual design effort involves locating the wireless access points (each access point, or AP, is a combination radio transmitter and wireless-to-Ethernet bridge) so that they are as far apart as possible to minimize interference and cost, while assuring they are close enough together to avoid gaps in the wireless coverage.

Equipment purchase includes the wireless AP and antenna. The cost of the AP depends on the type you choose to deploy. The standard "enterprise" type of AP typically deployed in a university or corporate building is less than $1,000. Small access points targeted for the home or small office are about one-third the price of the full-sized, full featured AP, while a device that supports authenticated connections and can serve users in more public settings may be twice the price of an "enterprise" AP. Beyond the choice of access point type, the number of access points a particular installation will require can vary greatly. And installation costs will include running power and data to each AP location.

Once the AP and associated outlets have been installed, those devices need to be connected to the campus LAN. One option is to connect the AP to the existing building network. This may be the most cost-effective way, but one must determine desired outcomes for the users of the wireless LAN. Because we wanted our end users to have seamless connectivity anywhere on campus (and as they traversed the campus), we chose to set up a parallel wired network and connect each AP to that wired LAN. We gave that network its own subnet of IP address space so that end users would not need to reboot or renew IP address leases as they roamed the campus. While this approach is more expensive, we felt it better met the current needs of the campus wireless users.

The amount of effort required to manage and support wireless will vary by institution. At Carnegie Mellon, the added effort to support wireless has been relatively modest, while managing the wireless network has been somewhat more labor-intensive.

Develop a Recommendation

Here are other considerations to take into account when formalizing a recommendation of whether to proceed with a WLAN deployment:

• Who are the major constituents and likely users? Will the various constituencies welcome the addition of a wireless network or will they view it as a distraction, an intrusion, or even a disruption?
• Is it important to cover the whole campus, or only a few key areas? Will users want to roam from location to location, remaining connected, or will they be satisfied with access in a few specific areas?
• Are there existing research activities that may create more support or demand for deploying wireless? Or would they or other potential sources of radio frequency interference make a wireless deployment particularly difficult?
• Is the technical and support staff necessary to deploy, maintain, and support a second network available? Are there sufficient ongoing capital funds to sustain a second network, with an expected refresh rate of every three to four years?
• What is the optimal location and time to begin experimenting with and/or designing a WLAN, given the requirement to access every office, room, classroom, and common space within a given design coverage area?

Despite their status as an emerging technology, wireless networks show great promise for enabling ubiquitous communication among user communities. So whether the final recommendation is to begin a deployment or not, keeping close track of new developments in wireless technologies will remain good advice for the foreseeable future.

Tracy Futhey is executive director of Computing Services at Carnegie Mellon University. (tf01@andrew.cmu.edu)