High Education Tackles The Problem Of Wi-Fi Capacity
Colleges are finding ways to meet the wireless computing demands of large concentrations of students--and businesses could learn a thing or two from them.
When I started college in the early 1990s, my network connection consisted of dial-up access to a serial terminal on the sole SCO Unix server, and the campus enjoyed a high-speed 56-Kbps dedicated circuit to the Internet. Fast-forward a decade or so: Big campuses have network access in dormitories, plus wireless overlays in public areas and classrooms. Most students own laptops, and they want their Wi-Fi.
Educational IT professionals worry primarily about capacity, while their enterprise colleagues are concerned about coverage. In offices where wired ports are still the mainstay, wireless access may need to span boardrooms, guest offices, and executive suites on different floors. But at universities, students congregate at libraries and large lecture halls. Rather than five laptops in a boardroom, think 40 students at study carrels using one wireless access point. This is an early look at what businesses will experience as their wireless trials and initial deployments morph into production networks, and they move toward broad use of wireless connections.
Haven't you learned? Access is everything.
Photo by Chris Hondros/Getty Images
Adding an access point within the ceiling won't solve capacity problems. First, there's the challenge of channel selection. Fred Archibald of the University of California at Berkeley's School of Electrical Engineering and Computer Science likens channel assignment to a Rubik's Cube--change it in one place, and the whole puzzle must be re-examined.
Deployments based on 802.11b/g can use only three nonoverlapping channels in the 2.4-GHz range. Adding an access point on channel 6 between two others on channels 1 and 11 may be OK. But if there are access points using channel 6 on the floors above and below, the resulting co-channel interference will reduce overall system performance in a moderately to heavily used environment.
To help solve this, Archibald's group takes advantage of Cisco Systems' Unified Wireless Network (formerly from Airespace), which uses 1100 Series thin access points with a management system that offers auto-channel selection. The team deployed one access point for every 3,000 square feet. Another option is to use 802.11a. The additional 255 MHz that the FCC allocated means there are essentially 23 or 24 nonoverlapping channels in the 5-GHz range.
This would resolve channel selection even in the densest wireless environments, but schools and businesses alike have demonstrated little interest in an option that adds to the cost of client cards and access points. A dual-band access point costs from $100 to $200 more than one that's 802.11b or 802.11g, though dual-band wireless cards are only $10 more and are becoming standard fare. This resistance will lessen as prices drop and IT groups face the realities of dense wireless deployments that require better performance.
Rein In Signals
Micro- or pico-cell deployments also tackle the capacity problem by regulating output power to limit signal propagation and potential co-channel interference. This is tricky if a minimum link rate of, say, 36 Mbps is required because the power must remain high enough to attain that rate.
Oakland University in Rochester, Mich., uses Proxim wireless 802.11b/g equipment in its wireless-only dormitories. Oakland IT initially scaled its access-point-to-user ratio to 25-to-1, but that's now down to 15-to-1, with high-complaint areas enjoying a 10-to-1 ratio; it seems the squeaky wheel gets the access points. It's not unusual for a wireless analyzer to pick up 15 to 20 access points at one time, because they remain at high power to provide adequate coverage despite the attenuation caused by brick walls and dorm-room furniture. UC Berkeley's Archibald cites a similar experience, even with Cisco's auto-power-control features. He sees average peak levels of 20 to 25 users per access point, with spikes as high as 40.
Additional problems arise out of micro- or pico-cell architectures. For example, many older client cards don't support proprietary standards such as Cisco Compatible Extensions, or CCX, that share metadata such as the number of clients the access point has for load-balancing purposes, radio-frequency information, power output levels, and more between clients and Cisco access points. Thus, they can't automatically turn down their output, resulting in power asymmetry.
The client card transmits far beyond its associated access point to distant clients and access points on the same channel, degrading overall wireless network performance. And while Cisco's CCX program includes about 95% of new wireless chipsets sold, only Cisco access points support CCX and can request that the CCX-compatible client turn down its output.
2014 Next-Gen WAN SurveyWhile 68% say demand for WAN bandwidth will increase, just 15% are in the process of bringing new services or more capacity online now. For 26%, cost is the problem. Enter vendors from Aryaka to Cisco to Pertino, all looking to use cloud to transform how IT delivers wide-area connectivity.
The UC Infrastructure TrapWorries about subpar networks tanking unified communications programs could be valid: Thirty-one percent of respondents have rolled capabilities out to less than 10% of users vs. 21% delivering UC to 76% or more. Is low uptake a result of strained infrastructures delivering poor performance?
Join us for a roundup of the top stories on InformationWeek.com for the week of December 14, 2014. Be here for the show and for the incredible Friday Afternoon Conversation that runs beside the program.