Sunday, June 12, 2011

Cisco is helping hotels block your mobile data

Just yesterday I got home from a trip to go present at the Float Mobile Learning Symposium in Peoria. It was relatively nearby, and I don't see another vacation in my future, so I hauled the whole family with me and we made it one of our typical road trips, avoiding interstates at all costs.

 Between them, someone was on the internet pretty much the whole trip there. And that was surprising, and pleasant. Even being all spread out, the US seems to be moving to a European/Scandinavian model of total access. The Clear aircard (falling back to Sprint, actually) was able to provide mostly rather good, and always passable internet. Really, 100% coverage, door to door. Verizon failed for the handsets, but oh well. I am sure they will get there also.

Yes, door to door. And when we get to the hotel, it failed. I spent a couple days ranting about how lame it was the internet failed in the largest city we'd been in since leaving home. I paid the $10 a day to get internet in the room (pretty lame internet at that), but then noticed that the Clearspot still worked in the car. In the parking lot. Of the hotel. Wait a minute... So, I did some tests.

  Click to download a PDF version of this chart

 The results seem clear. It's not loss of signal, so should not be (purposeful or not) passive interference. It is pretty unlikely to be accidental interference in my opinion. And I find it interesting that the only place I have stayed at in years that still charges for internet has this behavior. Hmm...

Details of the testing below. All speeds logged with speedtest.net Android application on Motorola Droid 2 Global (Verizon, in CDMA mode). Tests performed several times in each location, and averaged. Additional testing performed with Clear (operating on Sprint 3G network) aircard and Sprint handsets via testing software on computer, as well as other speed testing protocols. Not included in averages, but similar results observed.

Since these devices operate on different frequencies, that would seem to rule out interference as the culprit. I didn't bring any live SIMs, so couldn't try any of my GSM handsets. There was no corresponding change in SNR, received power or the handset’s transmit power (indicating unusual power management). Latency was quite low for a mobile network (around 120 ms) and consistent for all tests, at all times. For the handsets, no change in voice quality or the vocoder level was observed. All effects were purely in the speed of mobile network traffic.

Cisco AP unplugged – A Cisco device was found under a table in the guest room. It is well-secured to the table, so the back (with FCC ID and so on could not be found) but it appears to be a Cisco Aironet® 1130AG Series wireless access point (e.g. a WiFi router). When it was unplugged (it used a single Ethernet cable, presuming with PoE) the in-room speeds immediately doubled, to broadly the same speed as any other in-hotel area, outside of a guest room.
Hallways, common areas, conference center – Numerous locations were tested. Speeds were around 10% higher on floors with no guest rooms. This speed was encountered even in places such as a fire stair; with cast concrete construction, no windows and heavy doors, it would normally be the worst place to attempt to get a signal, but still offered broadly the same signal, and twice the network speed of the guest room with it’s broad windows.
Bridge – There is a conference center across the street, with a second floor pedestrian bridge of glass wall construction. This offered only a slight increase in speed, with significant fluctuations during the test. These fluctuations were not encountered in-room.
Outside hotel – All areas outside the hotel, in areas up to 1 mile from the hotel, had broadly the same coverage and speed, varying from 1100 kb/s behind walls where signal was masked, to 1,600 kb/s in open areas or high ground. This includes standing against the hotel. Tests were performed against the wall of a common area with nothing but windows, and immediately then was switched to outside the same location. Three feet away, with only glass in between, there was no change in SNR or received power, but an immediate tripling of traffic speed.

No comments: