Using EV-DO
My experience in using EV-DO to get ubiquitous connectivity.08-Apr-2005

For an overview of my experience read shorter comments on EV-DO as a Taste of Connectivity.

I don't want gadgets, I want opportunity. Even if I'm given a service I look beyond what it is supposed to be to what it can be and, alas, what it can't be. Or at least, cannot easily be. EV-DO means I can be connected everywhere (that matters) but it's not the same as what I can get at home. Bluetooth liberates me from wires but tethers me with faux wires. For me this is all wonderful but I fear being trapped in the present reality of these services rather than viewing them as simply hints of what is possible.

I need to start with a disclaimer -- I've already written about how Bluetooth has failed so I have a stake in its failure. Yet it seems to be succeeding. Same goes for using the cellular phone system for data rather than the Internet. I'm still sticking by my statement since the current seeming success of Bluetooth is based on its suitability for applications defined years ago. I'm probably one of the most avid users of Bluetooth because it's the only way I can get some of the capabilities I want.

This is frustrating because Bluetooth is backwards looking. New capabilities take a long time to get adopted by the committee process. If I really use Bluetooth I run into the 7 device limit so I needn't fear runaway success. Getting my Bluetooth GPS to work required dealing with the legacy of "COM" ports. It's as if all new Touch-Tone® (not really ® anymore) phones had to send pulses because all dialing had to conform to legacy approaches. These legacy concepts are endemic to Bluetooth. In connecting my HP 6315 to my Bluetooth access point I had to use a virtual null modem. Next I'll have to feed virtual oats to feed the horses that power my car?

I had to determine that my GPS was connected on COM11 and then type it into my application. Even that wasn't enough for the Delorme GPS. After calling customer support I was told I had to update my Bluetooth drivers. I naively assume that by using both the Toshiba and the Microsoft update programs I would have had the latest version. Wrong. While I could use the Socket (resold by Copilot) Bluetooth GPS, the Delorme one required the new Bluetooth "stacks".

The term "stacks" is interesting -- it is a term for the layered dependency and the user must be careful to assure the proper software at each layer but installation woes are another topic I want to write about.

The whole COM port thing is a kludge to make things work despite Bluetooth. Rather than having a standard TCP stream with the two ends finding each other, I have to sit in the middle and stitch together virtual wires to make my applications work!

EV-DO comes from the CellCos that brought you Bluetooth. Here again I have my qualms because I don't want the carriers and their scarcity-creating protocols sitting between me and the rest of the Internet. $80/month for a DSL-speed connection ain't cheap and it isn't really as good as DSL. Verizon promises a .5 to 2mbps connection. Actually they don't promise anything but simply say that I can get a burst speed up to 2mbps.

IEEE Communications Magazine April 2005 issue features cdma2000, the family of services which includes EV-DO (actually 1xEV-DO Rev A). Even though "Communications" is a technical publication, the key measure is ARPU or Average Return per User -- a revenue measure! That's understandable but it leads to problematic design decisions. The reason why IP networks triumph is that they are not hobbled by revenue-defined architectures. In the protocol diagrams IP is above PPP which is above a whole set of lower layers thus losing the efficiency of best efforts opportunistic delivery. The Aloha Net was simple packet radio net -- send a packet. If it failed, try again. That basic design begat the Ethernet and inspired the Internet.

Landline telephony is converted to IP just to stay competitive; the cellular networks will face these economic pressures packet connectivity becomes more available as people share their access and as it becomes a municipal utility.

In the meantime, EV-DO is useful because the carriers have invested a lot of capital in building out their system and we still treat wireless communications as exclusive property thus making them the only, at least the most comprehensive, game in town in terms of coverage.

Given that, it's no surprise that EV-DO is very good compared with the alternatives -- slower speed over cellular or hot spots intended for web browsing rather than causal connectivity. At an airport I can be on the plane (on the ground with the doors open for the superstitious) and be online whereas airport Wi-Fi coverage rarely extends that far and demands a payment. Sure, it's faster but for now speed is less of an issue than availability.

Attending F2C I found that the EV-DO connection was slightly faster than the connection I got at the hotel. The hotel claimed 10mbps but I got a little less than 1mbps for downloads. When I did the same download with EV-DO it was a tad faster. Perhaps that's why they gave away the connection as an amenity instead of charging for it.

While in practice the EV-DO speed isn't too bad but still far less than the 20mbps connection I shared at F2C. While I haven't done full testing, I do notice long latency times for EV-DO. PC Magazine's testing showed delays of .5 to 1 second. This isn't too bad if you are downloading or, perhaps, if you are streaming with a large buffer. It's a higher speed version of 1XRTT that Verizon offered -- it was "up to" .15mbps. While Skype does work over EV-DO (and 1XRTT) the delay is problematic.

In testing from home I see the VZAccess meter showing a maximum of 457kbps down and 147kbps up which is far better than modem speeds but not impressive by comparison to standard high speed connections. The results of http://www.broadbandreports.com/speedtest weren't quite as good. One test reported 112/93. Latency is a major factor (though speed meters generally don't report it) so performance tests should be done comparing small files to large files. Using a 2MB file (AKA 16 megabits) my upload got aborted but the download was at 160kbps. This is actually far lower than the times I get when testing the service at my hotel in DC. This is indeed a case of "your results may vary". In retesting the up speed, I found that 1 megabyte works but 8 megabytes causes the connection to abort -- hmm …

Compared with using a dial up connection, the speed of EV-DO is refreshing and wonderful. But it isn't the same thing as getting the high speed symmetric I want to really participate in the Internet. I should be able to not just watch VDTV (Very High Definition TV -- above the piddling HDTV 1080p) but also originate content and have a high resolution conversation. Once again we have an example of a capability which is great if we look backwards but barely so. It is the enemy of the future because the architecture is not suitable to providing far more abundant capacity.

The $80/month price is also very high. I could save money if, instead, I just added the EV-DO capacity to my cell phone and used it as a relay to the rest of the Internet. In fact, that's what I do now using 1xRTT on my Samsung. That only costs $45/month for some reason even though it is the identical service. How can Verizon get away with such price discrimination? True, I'm also paying for voice service on my cell phone but that was true when I had a 1XRTT card for my PC and could use it to make phone calls.

I was planning to replace my i700 with an i730. The price for EV-DO is the same as for 1xRTT. This is similar to DSL which costs the same whether I get full speed or half speed. I actually get the latter because Verizon has a policy of looking the length of the wire to determine the maximum speed and refuse to test to see the actual speed -- obviously customers are a burden so no need to try at all, let alone try harder.

Since the new i730 isn't yet available from Verizon (even though Samsung has had it available for months) I decided to try the card even at a premium price. I then realized that Bluetooth has met its match with EV-DO. Its maximum speed is below EV-DO speeds. That's true even for the proposed Bluetooth 2! This kind of cripples the idea that one should use ones cell phone and Bluetooth for Internet access. That's the price of locking failures of imagination into silicon!

It's unclear how Verizon will cripple the i730 but I presume they will leave it capable of doing 802.11 (Wi-Fi) so I still have the option of using a simple application to share the connection. After all, the i730 is a pretty powerful computer in its own right -- too bad we malign it by calling it just a phone or a PDA.

Assuming I can use my phone as a relay point it makes little sense to pay $80/month to connect my PDA if I can easily connect via my phone for $45/month. I presume EV-DO and the voice path can use the same transmitter. Since we are communicating using packets then the voice is another data type over the network there's not reason to have to choose voice or data -- they are just packets. But I expect that Verizon will have make a policy decision disguised as a technical necessity to force me to choose one or the other.

One other problem with using the phone as my network relay is that it has a smaller battery than my laptop but this shouldn't be an issue now that we have a universal power supply -- see my comments on "USB Power!"

To repeat the point I made in EV-DO as a Taste of Connectivity: I am an enthusiastic user of both EV-DO and Bluetooth. I want to encourage others to take advantage of the technology but I am cautious -- people tend to get a stake in an existing technology and resist the newer technologies even if they are much better and we already know that there are better technologies. We've already seen landline connectivity transformed and telephony itself becoming just a minor software application. 802.11 allows for new applications while Bluetooth is stuck in the past. EV-DO and Bluetooth are the best of the past. To go further we'll need to shift to native IP for wireless connectivity just like we are doing for wired connectivity.