Skip to content Skip to navigation

LCSR Wireless Experiences

or

I am Livin' on Channel Z!

(by Charles McGrew)

Impetus We started getting requests from our users that they be allowed to attach their laptops (either personally purchased or grant-purchased) to the Rutgers network. The drawbacks to this boil down to: if we put hard network connections (10baseT jacks) in labs and/or public places, we cannot control who actually attaches to them.

Solutions involving switched ports and hard-coding MAC addresses into gateways seemed going in the wrong direction; also, soon enough the ratio of laptops/palmtops to users will become 1:1; the hard-wired approach involves a massive increase in the number of connections in labs and public rooms, with a simultaneous rise in cost and risk of network penetration by naughty people.

The Wireless Approach If we instead used wireless hubs to provide network connectivity, all the hardware costs go 'way down. (The drawback is that users need to obtain a wireless card -- pcm/cia wireless cards are less than $150 -- $119 today according to www.pricewatch.com, we thought this wouldn't be too big an imposition, besides, the cards have uses for the users outside of Rutgers.) The networking hardware (for the hub-side of things) can be secured away from user areas, the cables connecting the wireless hubs to the Rutgers network are secure, and access to the hubs themselves can be limited by MAC address (which can be matched 1-to-1 with a user).

Further, if the wireless hubs all claim the same wireless domain, and if a dhcp server assigns IP addresses on a 1-for-1 basis, the machine-to-IP correspondance is maintained (so we can track evil doings), and the users can use their machines anywhere in the radio covered area -- even moving from place to place and maintain TCP connections (that is, telnet won't die if you pick up and walk to another room, or even another building -- so long as you don't pass though any "dead" areas; the clients automatically switch base-stations if they discover a better one.)

Our plan Most of the work for this has been done by Hanz Makmur and Doug Motto, and me. We tried to guage where wireless hubs (Apple Airports) could be placed to give good (at least 5.5Mb/s, but mostly 11Mb/s, the maximum for hub) coverage for our user population, with special attention to Professors, PTLs (Part Time Lecturers), and TAs, since they are our main current candidates (they use laptops for class presentations) -- plus the Core Lobby, 1st floor Auditorium and the 3d floor conference room.

We came up a layout for 8 airports to cover 1st, 2nd and 3d floor of CoRE, and 3d and 4th floor of Hill Center -- experiementation showed that we really need 6 for Core (two corners and center-of-building) . All the airports sit on the same subnet (128.6.157 -- lower half), and have a single dhcp server to assign IP's (knowfuture.rutgers.edu - the dhcp server being set up by Ken Harris). Additional airports will be placed in the "cereal hall" -- hill 2nd floor, and there is one already placed in a wiring closet on the first floor of core (for the lobby and auditorium).

A note about physical security: we tested and obtained a remarkably inexpensive Tripwire Shockpad(TM) for each Airport mount point. We have certainly learned the value of good insulation while installing them!

Experience So Far The images below detail the wireless coverage sampling done so far.
DB values for Hill and Core detailed below describe the "signal strength" value reported by Ianoco-based software on a Dell laptop. Decibel values less than 10 are unacceptable, and generally the higher the better.

Intuitive Results So far we've determined that elevators really kill reception; if there's an elevator between you and the airport, you lose. (Since elevators are big Faraday cages, this isn't so surprising.) Also, coverage may be obtained from airports on other floors -- though the range of an airport on another floor isn't anywhere near as good as one on the floor you're on.

Further, coverage in hallways is generally very good, but signal strength can drop 30% or more when entering a room (either in Hill or CoRE), and is generally worst in the center of the room, and good just inside the door or over by the office's window (for offices that have windows.) The halls generally act as a "wave guide"; the physical layout of the floors has a lot to do with reception. (Hill, with its 'S' shape and two sets of elevators in the center of the building proved the weirder of the two.)

An AirPort's physical orientation is a big factor in coverage. Airports send/receive significantly better out the sides than out the top/bottom. Care must be taken to achieve the coverage configuration you want.

Core 2
General Coverage values
Coverage and the Airport it comes from
Core 3
General Coverage values
Coverage and the Airport it comes from

To see coverage by airport, look here


Hill 3

Hill 4

To see coverage by airport, look here.


Update: Since this document was originally written, we have added wireless base stations (of various manufacturers) in Hill Center, on the first floor (over by 114/116 -- the main class rooms on that floor), and on the second floor (248).

Additionally, we have created a web-based authentication tool that allows users with a valid Rutgers account (DCIS, Eden, RCI, Camden, Newark) to authenticate themselves to use the wireless network. The original scheme to update allowed MAC addresses has been supplanted by a much more flexible scheme that allows updates to authenticated user machines automatically (and also automatically removes them from the authenticated list.)

We have done testing with the Lucent external antennas that show that they give a 5-10db improvement in some (but not all) situations. Generally, it is far easier to add a base station to improve coverage than force users to purchase the extra antenna.

The Future The advantages of the 'Roaming Network' approach go beyond our own convience and network security; if this scheme were to grow outside of just a couple of buildings -- to say, classrooms -- the advantage for teachers and students would be enourmous. A pretty cheap way to get to "smart classrooms", I'd say...


Appendix A: Airport Channel Selections

This is the usage of Airport channels used by the DCS/LCSR Wireless project:

Airport
Name
Location
(Or nearest room)
Channel
Airport-1 Hill 488 7
Airport-2 Core 227 5
Airport-3 Hill 376 9
Airport-4 Core 1st 9
Airport-5 Core 324 3
Airport-6 Core 302 1
Airport-7 Hill 427 1
Airport-8 Core 338 11
Airport-9 Hill 248 5
Airport-10 Hill 413 11
Airport-11 Core 211 7
Airport-12 Psychology 7
Airport-14 Hill 116 3
Airport-15 Core Auditorium 3
Airport-16 Hill 114 7
smc17 Core 116 1
smc18 Core 234 3
smc19 Core 102 5
smc20 Core 201 11

Sorted by Channel:
Channel Airport
1 Airport-6(Core 302), Airport-7(Hill 427), SMC17 (Core 116)
3 Aiport-5 (Core 324), Airport14 (Hill 116), Airport-15 (Core Aud.), SMC18 (Core 234)
5 Airport-9(Hill 248), SMC19 (Core 102)
7 Airport-1(Hill 448), Airport-11(Core 211), Airport-12 (Psychology), Airport-16 (Hill 114)
9 Airport-3(Hill 376), Airport-4(Core 1st)
11 Airport-8(Core 338), SMC20 (Core 201)
13 Airport-2(Core 227), Airport-10(Hill 413)