Brian's Brain

Last weekend while in Las Vegas, I noticed promotional signs at each Monorail station advertising Linksys' WRT54G3G-ST Wireless-G Router for Mobile Broadband. The company originally unveiled the product in April of 2006, and at that time I expressed my interest in reviewing it as a competitor to Kyocera's EV-DO and 1xRTT-sharing KR1. A reminder email to Linksys' PR team promptly resulted in a WRT54G3G-ST arriving at my front door yesterday, along with a Sprint-supplied Novatel Wireless Merlin S720 EV-DO Rev. A PC Card.

I've long been interested in evaluated out EV-DO's Rev. A version, which upgrades the EV-DO rev. 0 predecessor that I've extensively tested and long used. And Sprint just recently turned on 'A' service in the Sacramento area. So while I haven't yet tested out the Wireless-G Router intermediary, last night I installed the PC Card software on my laptop and, after calling up Sprint to activate service (which only took around a minute, even though the telephone rep indicated that provisioning might take up to four hours), I was online and surfing away.

If you look at Wikipidia's definition of EV-DO, for example, you'll notice two particular documented advancements of EV-Do Rev. A versus Rev. 0: decreased 'ping to ack' latencies and increased upload speed. To verify or refute these claims, I resurrected the same set of elementary tests I've long used when evaluating wireless and other LAN/WAN technologies:

• A 'ping' loop to www.yahoo.com, and
• Bandwidth testing at BroadbandReports.com, specifically using the Sun Java plugin option and to MegaPath Networks' San Francisco server

If you go back and look at any of my past evaluations of Sprint and Verizon's EV-DO rev. 0 service, you'll see that 'ping' times are on the order of 250 msec. Last night, the pings using my Sprint-based Sierra Wireless AirCard 580 again averaged that same latency. Pings to Yahoo's URL with the EV-DO rev. A card, on the other hand, were 'ack'd' by Yahoo's server noticeably faster, on the order of 160 msec. These delays are still unacceptably long for VoIP communications.... then again Sprint doesn't necessarily want you to run third-party VoIP over its cellular data network anyway, right?

I then ran the BroadbandReports.com bandwidth test, three times each on the EV-DO Rev. 0 and Rev. A cards. Here's the Rev. 0 data:

And here's the Rev. A data:

Notice that upload speeds in particular are significantly faster in the Rev. A case.

Verizon also recently sent me an EV-DO Rev. A card, the Sierra Wireless AirCard 595, and I already own the Verizon-served EV-DO Rev. 0 Kyocera KPC650. Verizon hasn't yet publicly launched Rev. A service in Sacramento, and my testing last night confirmed that the AirCard 595 delivered only backwards-compatible EV-DO Rev. 0 speeds here (at least when connected to the closest nearby Verizon tower). However, I'm headed to Silicon Valley (where both Sprint and Verizon have active Rev. A service) for meetings in two hour's time; I'll bring the laptop and all four EV-DO cards with me, and I'll report back any interesting developments I find.

My pending trip will also enable me to test EV-DO Rev. A performance not only while I'm static but also while the Amtrak train I'm on is moving at high speed. And, when I get back to Sacramento, I'll pair the Sprint card with the Linksys router and see if the latter causes any incremental performance degradation, as well as sharing any other thoughts I might have on the WRT54G3G-ST. Stay tuned, and until then, share your thoughts on the data I've already obtained!