Wireless comparison: 802.xx vs. EVDO a test and a future prediction

It's the Suits versus the Cowboys in the battle for your wireless future. The Suits are the cellular carriers; the Cowboys are entrepreneurs implementing the Institute of Electrical and Electronics Engineers' 802.xx protocols. The Suits bring zillions of dollars in wireless infrastructure, a long operating history, a huge base of captive customers, thousands of roaming agreements and vertical integration of systems, software and services. The Cowboys bring ... uh, they bring ... uh ... well, they'll improvise. My money is on the Cowboys. The battle replays minicomputers and PCs. The minicomputers were the Suits. The PCs were the Cowboys. PCs won handily. Here's the story for wireless.

My office and my house are in the middle of nowhere. I'm at the end of 1.7 miles of pavement, on a ridge that sticks out between a couple of taller ridges. There are good views from my property, but there's no clear view to geosynchronous satellites, to heavily populated areas, or to ridges with repeaters. There's no cable TV and no phone company central office within five miles. Around here, fiber is a word on cereal boxes.

This is not a good situation for a technology analyst in Silicon Valley, so I have been looking for anything that might improve things. Several years ago, I got a WebRamp box. One interface to the WebRamp is three modem connections; the other is Ethernet. I could attach one to three 56 kilobits per second (Kbps) modems to separate phone lines, plug them into the WebRamp, and share the aggregate bandwidth on my home network over Ethernet. The WebRamp managed the connection sharing to the Internet service provider.

I limped along like this for several years. WebRamp, which had been a startup, went public, was bought and disappeared. That was the end of software updates; performance degraded.

Then Verizon announced EV-DO, Qualcomm's Evolutionary Data Only system, and began turning it on in California. With EV-DO, one purchases a PC card from Verizon that provides data service over the cellular network at promised download speeds of 400 to 700 Kbps. Hey, these rates qualify as broadband in the U.S. and in third-world countries, and they are wireless and mobile. In any case, they're way better than the 26 Kbps of my analog phone line. At the beginning, the price for Verizon's "unlimited broadband access" was $90 a month, but it has since dropped to $60. The cost was high compared to the phone line, but it offered about 20 times the data rate.

When I visited a Verizon store, I had three questions:

1. Is there EV-DO coverage in my area

2. Will the PC card and its software allow me to share the connection on my home network

3. Does the service agreement allow me to share the connection on my home network

________________________________________

Verizon's in-store representative had detailed maps to answer the first question. According to the company's coverage maps, there was no service in my area. But based on the one bar of power on my Verizon handset, I believed the company might well be wrong.

I tried asking my second question about six different ways, but it didn't work. I knew that by using a Windows operating system program called Internet Connection Sharing, the laptop could share its connection with the rest of my home network--that is, unless Verizon's software interfered with Windows. For each phrasing of the question I got a puzzled look and the same answer: The EV-DO card plugs into the PC card slot, so it can't be plugged into a desktop at the same time. OK, no answer to the second question either.

The answer to my third question was also not known in Verizon's outpost. I was zero for three, but desperate, so I signed the two-year contract and took the experiment home. I would have 15 days to answer the questions myself before the contract became binding.

The Experiment

I installed the Verizon software on my laptop computer and plugged in the EV-DO card. The software recognized the card, configured it, and established a connection to the cellular network flawlessly and quickly. Impressive. With a single bar of coverage, I could access the Internet through the cellular network from my home! The coverage maps were wrong; as I had suspected, the connection-mapping explorers hadn't been to my neighborhood. The answer to question one is yes, there is EV-DO coverage in my neighborhood; one down and two to go.

My home network has a server, a PC-based PBX (private business exchange), two desktops, two laptops, three wireless access points and several experimental computers attached to a common Ethernet network.

I installed the Verizon software on three laptops and tried the EV-DO card in each of them. Nothing worked. Verizon's software isolates the laptop that has the card. It's not even possible to access files stored on other computers while the Internet connection is active--the Verizon software offers only a black-and-white choice: connection to the cellular network or connection to the local network.

That's the end of the experiment. I cannot use the EV-DO card as my home network's Internet connection. Even the expensive route of using one card for each computer doesn't work, because I regularly ship files around the network. The agreement does seem to say something like one computer at a time. Why would Verizon do this

Verizon has "broadband" coverage in my area that no one else offers. It is priced well above much faster cable services that don't care how many computers are connected to the network. Why does Verizon insist on controlling how I use the card

________________________________________

A few years ago, it looked as if all the advantages in the competition for wireless Internet access belonged to the cellular networks. They could incrementally upgrade their networks and their services, from basic phone service to data-based services to multimedia. They had great connections to the backbone network. Their already-in-place systems could even support mobile connections. And they had the accounting systems to make billing simple for the customer.

Contrast that position with the state of Wi-Fi. Coverage was sparse and mobile connections weren't even dreamed about. Companies were searching for a business model that would make Wi-Fi services worth offering at all--forget universal coverage or mobile access.

The battle for wireless broadband access looked like a lopsided match between cellular networks, with a host of advantages, and Wi-Fi networks, still searching for a business model.

The Answer

My experiment with Verizon's EV-DO slapped me in the face with the answer to who will win: cellular or 802 Successful companies develop a culture that suits the competitive situation. For example, Intel's culture of intense focus on next-generation microprocessor design and on leading-edge semiconductor processes led it to the top of the highly competitive microprocessor market. Cellular carriers have cultures expressing three primary "genes" in the DNA of telephone companies: a build-out mentality, vertical integration and complicated pricing. The build-out mentality comes from the legacy of the government's treatment of the electromagnetic spectrum as physical property, encouraging carriers to build out networks to the extent of their spectrum ownership. Telephone companies are vertically integrated in providing the network infrastructure, the services and the customers' devices. Finally, complicated pricing helps to obscure profits in a highly regulated business that includes price controls.

But the culture that builds momentum for a growing company may paralyze it when things change. Look what happened to phone companies after deregulation. They had cultures suited to regulated-monopoly environments. Similar problems befall government-contracting companies when they move into commercial markets. The "Wild West" nature of IEEE 802 systems is changing the competitive environment for the cellular networks. As my experiment with Verizon's PC card demonstrates, companies that developed in a regulated environment will find it difficult to adapt. I concluded that the inertia of the cellular networks' culture will prevent them from exploiting their huge initial advantages; they will lose to the 802 zoo.

Verizon integrates the customer device (the network-access hardware and software), the cellular network and the service. This integration is part of its competitive problem. The market should have competing end-user device hardware and software; it should have competing cellular networks; and it should have competing services. Imagine how much better each would be if provided by separate competing companies. You don't have to imagine it; that's what's happening in 802. The 802 world is like the PC world, horizontally fragmented, not vertically integrated.

But before finally giving up on the cellular players, I decided to give them one more chance.

________________________________________

A few months after my experiment with Verizon's EV-DO card, I decided to experiment with a similar card from Cingular. As a prelude to the trip to town to visit a Cingular store, I checked the Internet for the latest options. I found the Kyocera KPC650 PC card and a companion Kyocera KR1 "mobile router." The KPC650 is an EV-DO card that is an option with Verizon's wireless broadband service. The KR1 mobile router is just what it sounds like--and exactly what I needed. It is a wireless access point with four Ethernet ports and a slot for a wireless PC card, such as the KPC650. I headed to Fry's Electronics to buy one. At the store, I found D-Link's DIR-450 EV-DO-compatible mobile router. The store had only one; I bought it and headed for a Verizon store. Since I had already experimented with Verizon's wireless broadband, I knew I had reception at the house. (No need to start the experiment over with Cingular's service.) So I was back with Verizon.

I inserted the KPC650 and plugged in D-Link's mobile router. Nothing. I installed Verizon's software in a laptop and initialized the KPC650. It worked perfectly. I popped the card out of the laptop and inserted it in the mobile router; this time it connected flawlessly. I uninstalled the broadband-access software from the laptops. I established a wireless connection between the mobile router and a laptop and configured the router. I connected an Ethernet cable from the mobile router to a newly installed Ethernet card in the server and configured the server to forward packets from the local network to the mobile router. That extended Internet access to the wired home network and to the rest of the wireless access points on the local network through the server's firewall. (The mobile router also has a firewall.)

I'm happy. My Internet connection is at least 15 times faster. Two weeks later, I installed a high-gain antenna to replace the miniature antenna that's on the KPC650. Testing the new antenna, I found that although it does not enhance the top data rate, the robustness of the signal improves. I canceled my dial-up service and returned two copper connections to the phone company. With the three cancellations, I'll break about even for the cost of the new service. I have gone from four copper phone lines to one, so I no longer need the PBX to manage options among copper connections--another potential decommissioning ceremony and more savings.

The cellular carriers aren't dead. The mobile router should be popular with emergency services. Installed in a service van, fire truck or ambulance, it acts as a Wi-Fi access point, providing wireless Internet access for computers, laptops and handheld devices that are in or close to the vehicle.

The Future

In the 802 vision, you wear an earpiece connected by Bluetooth (802.15.3) to a handset that has a secure, mobile Wi-Fi (802.11) local-area connection to an access point and has ZigBee (802.15.4) connections to local-area sensors. The access point has a WiMAX (802.16) wide-area connection to the "lambda rail." The WiMAX base station supports 802.16e, so the backend of your car's Wi-Fi access point can connect to the WiMAX base station as you cruise the Interstate. If you cannot connect to a free-access network, your gear will connect to a cellular network.

Three developments may change this vision: cheap radios, mesh networking and smart antennas.

Systems from startup Mesh Dynamics contain three or four radios based on Atheros chips. The system is 2x6x8 inches, dissipates 400 milliwatts (mW) per radio and can be powered from an Ethernet cable (no need for an electrical outlet). One radio, operating 802.11b/g Wi-Fi with an omnidirectional antenna, serves as the access point for local connections. Two radios, operating 802.11a, form backbone link connections. The network organizes itself, establishing connections, routing around failed nodes and changing link channels as necessary to avoid interference. Radios, at $40, are cheap enough that two radios can be dedicated to the backhaul links.

Besetting mesh radio networks are four major issues: Interference between users' access links and mesh links; traffic jams through nodes close to the backbone link; single-radio scaling; and unacceptable delays through multiple hops. Using cheap radios, multiple-radio nodes, mesh networks and directional antennas, however, Wi-Fi networks already in the field demonstrate link distances and data rates that are as good as WiMAX will be able to provide--and they do it today at costs below what WiMAX will be able to achieve. Wi-Fi's MIMO systems already exceed the maximum data rates projected for WiMAX systems. I'm sure WiMAX's advocates will be able to point to advantages for WiMAX systems, but time and system cost are on the side of Wi-Fi. When WiMAX radios are cheap enough, it would be a simple thing for Mesh Dynamics to substitute WiMAX radios for the Wi-Fi radios that do the backhaul in its current systems.

Cities and even counties are enveloping their areas in Wi-Fi "clouds." One incentive is financial. It gives city or county employees anywhere, anytime access to information relevant to their jobs. Rather than blanket the city in separate, sparse radio networks for each service (fire, police, utilities, roads, etc.) the government consolidates infrastructure costs and provides better coverage.

For example, using systems from Mesh Dynamics, Sandoval County in New Mexico, an area a little smaller than the state of Connecticut, is providing megabit wireless access throughout the county. Complete, plug-in Wi-Fi radio modules cost only $40 to $100, depending on power output. Couple radios to directional antennas mounted on towers and Wi-Fi can span distances of 20 to 50 miles (in favorable terrain) for backbone communication at data rates of 24 Mbps (at maximum distance) to 54 Mbps (below 20 miles).

It is looking like Wi-Fi has the potential to provide universal access in cities and in rural areas, and it can also act as the wireless backhaul network. WiMAX will have an uphill struggle to achieve affordability before being overwhelmed by Wi-Fi deployments. Wi-Fi builds for volume, and WiMAX builds for performance. We have seen this before in the battle between PCs and workstations. Volume wins.

________________________________________

The main event, however, is competition between the vertically integrated telecoms that own licensed spectrum and the horizontally fragmented 802 zoo operating in unlicensed spectrum. It comes down, as I said at the beginning, to the Suits versus the Cowboys. The Suits have a head start: massive infrastructure, captive customers, exclusively owned spectrum and vertical integration for soup-to-nuts solutions. Undermining these advances is a monopoly-oriented corporate culture, proprietary solutions and the inertia inherent in vertical integration.

The Cowboys have open standards and a Wild West entrepreneurial environment. The Cowboys suffer from many competitors fragmenting the market and from operating in unlicensed spectrum.

My experience with Verizon's PC card and with a host of Wi-Fi products tells me that the Cowboys have the advantage. The vertically integrated cellular providers are handicapped by a corporate culture that makes them want to control the service as if it were the old switched-circuit voice network--and that will be unacceptable to generations of users that grew up with the wide-open development of the PC.

A key to the outcome will be how fast the Suits can learn from the Cowboys.

Excerpted from the December 2006 issue of the Gilder Technology Report.