For a technology that has been around for most of this century, it is hard to pin down exactly what is wireless data and where it is going. In its purest form, wireless data uses some form of the electro-magnetic spectrum (radio, infrared, or microwave) to communicate between two or more devices. However, this covers many different uses and applications: everything from cellular telephone modems to pagers, dedicated data terminals in moving vehicles to handheld radio modems, and new devices with built-in radios such as Motorola's (Schaumburg, IL) Envoy. Even mainstream computing vendors such as IBM (Armonk, NY) and Dell (Austin, TX) have latched on to wireless as their rallying cry to boost interest, sales and hopefully, profits in their wares. The next twelve months promises more of the same. What complicates matters is that the market is changing quickly as new players arrive with new products and service offerings, and that the relationships among the players is complex and require lots of study. This makes wireless data products harder to evaluate and categorize than wired data solutions. Dealing with all this new-found mobility that wireless data offers is one issue: since users are not often found at fixed locations, it is hard to assemble all the gear needed to go wireless, and harder still to track problems and evaluate the technologies without extended efforts. Another issue is that wireless data lies at the cusp of computing and communications: is a wireless transceiver more of a replacement for wired modems or local area network (LAN) connections? Not helping matters any is that an entire new vocabulary is needed to deal with this emerging world. Software developers that indicated their products were "dialing the phone" (when was the last time your phone had a dial anyway) and going "off hook" to make a data "call" or on a particular "telephone number" no longer make sense in the wireless world. There are no wires, no phones, no phone numbers, and no hooks. And finally wireless data can be still more of a stunt than a practical and mature technology: cutting through the hype and fluff will take a determined buyer or researcher to figure out the limits and business benefits of each particular wireless niche. In this last respect, the wireless data of the 90s has a long history: a group of Union College students crafted the first wireless baby carriage -- complete with antenna, batteries, tuner and baby -- 73 years ago this month in Schenectady, N.Y. (The city was home to a large General Electric plant and the college had one of the first broadcast AM radio stations, WRUC.) No evidence was recorded of any harmful effects caused by the radio apparatus to the baby, however. [To get rights to this photo, give Peter Blankman a call at 518 370 6000 in the alumni magazine office.] So where to start? Who are the major players? What's the overall market potential, and where is this industry going? Will there be rapid growth on a variety of fronts, or will many dollars be lost chasing dreams and few customers? And what are the actual realities of wireless technologies anyway? To answer these questions, let's first break down the entire wireless market into the following six categories, presented in no particular order, and assess the issues for each category: 1. Analog cellular modems. These include both standard data modems as well as fax modems for cell phones. (The modems themselves aren't wireless: they still need the radio inside the cell phone to communicate wirelessly.) BIS Strategic Decisions (Norwell, MA) estimates that there are 200,000 cellular customers that are using data modems, with an estimated $140 million worth of data transmitted last year, or about $5 a month of data calls. That doesn't make much sense: either most cellular modems aren't being used all that much, or there are fewer actual customers if you believe the overall total market value number. If this estimate is true, it is probably the largest current market for wireless data services both in terms of total customers and revenues. It is a rich market as well, with a variety of companies active and many available products. Leading vendors include AT&T Paradyne (Largo, FL), Data Race (San Antonio, TX) and Hayes Microcomputer Products (Norcross, GA). This last year saw several new cellular modems which fit into special slots on a variety of PC laptops. These slots, called Personal Computer Memory Card International Association (PCMCIA) slots or PC card slots, are about the size of a credit card and less than half an inch thick. The modems cost $300 or more, but prices are dropping fast as there is lots of competition and over a dozen manufacturers currently. They draw power from the laptop itself, which tends to limit their useful life to about an hour or so of actual communications. There are other cellular modems that are larger than PCMCIA cards as well for those laptops that don't have the slots, such as any of Apple's Powerbooks. But cellular modems have several issues: nationwide roaming, attachment cost, and reliable connections. Roaming is particularly difficult, as only a few cellular systems have made things easier for customers who cross coverage boundaries have found out. Attachment is cumbersome and costly, as no two cell phones have the same interface or plug that can be used to connect them to a laptop computer. Unlike ordinary wired telephones, cell phones lack the RJ11 jacks and have specialized electronics that make plug-and-play impossible with a wide variety of modems. For example, AT&T Paradyne's KeepInTouch modem only works with the Nokia (Sausalito, CA) cell phone. Typical prices for the connecting cables alone start at $100 and can be as high as $400. And reliable connections are still troublesome as cellular signals get dropped or fade as users move about while communicating. Data Race is one company that seems to be on the leading edge of cell modems. It offers adapter cables for the widest range of phones, including models from AT&T, OKI, NEC, Motorola and Pioneer. The adapters sell for less than $200 (which is still pricey given that you are only getting a cable) and connect the phone to its PCMCIA card modem called RediCard. While we have not tested the card with any cellular phone, this approach has lots of merit. Nevertheless, these obstacles are not insurmountable, as evidenced by United Parcel Service, the most notorious and visible cellular data customer. Last year, UPS finished outfitting all of its 50,000 delivery vehicles with cellular data modems. They hammered through agreements with a smattering of cellular providers to ensure nationwide coverage for their package-tracking system. UPS representatives report that the system works and works well. At the heart of the UPS system is a technology called a cellular modem pool. Essentially the same concept as a wired modem pool, a cellular modem connects to this apparatus (located at each carrier's switching center) and the data is then transported over ordinary land lines back to UPS' host computers. The modem pool is maintained by the cellular provider, and further ensures that no modem mismatch will occur between originating and destination modems. Ameritech (Hoffman Estates, IL), GTE (Atlanta, GA), and Nynex (Orangeburg, NY) have begun offering these modem pools to their commercial customers in many parts of their service territories, usually without any additional monthly fees above those for making the data calls themselves. Another interesting development is a product called Simon, jointly marketed by Bell South (Atlanta, GA) and IBM. It looks like a cell phone and behaves like one, with some additional features such as a cc:Mail client and fax capabilities built-in. Simon just started shipping last month and we were unable to obtain one for evaluation prior to publication. The cost of the device is $900 and it will work for eight hours from its battery, according to the announced claims. All of these signs point towards more activity in the analog cellular data arena: there are almost 15 million cell phones moving about the countryside and the potential market opportunity is huge: Users use the same applications, understand how wired modems work, and are already carrying PC-based laptops with them when on the road. If cellular carriers can reduce the cost and hassle of the modem connection, they should reap a windfall. 2. One-way radio receivers. Pagers are the most noticeable devices here. This category also includes a variety of other dispatch-related display terminals as well. About 15 million pagers, according to various estimates, are in current use. However, the vast majority of these devices have no displays ("beepers") or have one-line displays with sparse memory that can receive and store very little data. Newer and more capable products have been announced over the past few months, including two pager products that are small enough to fit into a standardized PCMCIA card slot in several laptop and palmtop computers. They can be found on Apple's (Cupertino, CA) Newton, Hewlett Packard's (Corvallis, OR) 100LX and Tandy's (Ft. Worth, TX) Zoomer palmtop computers. However, while the slots on all three palmtops may look the same, they require a variety of systems software so that a pager card designed for one computer may not work in another. For example, Motorola's Embarc News Card (Boynton Beach, FL) will work in the Tandy but not the others, and Wireless Access's AccessCard (San Jose, CA) will work in the HP LX but not the others. These devices cost anywhere from $50 to $500, the higher end of the scale occupied by the newer and smaller PC card pagers. Monthly charges start at $15 and quickly go to $100 depending on certain premium services selected. Batteries -- typically the AA or AAA commonly found in any airport newsstand -- last days or weeks. There are three critical issues before these advanced pagers will go "platinum" and sell in the millions of units: First, what is the business benefit over ordinary alphanumeric pagers? The PC card pagers require a palmtop costing several hundred dollars. They are more cumbersome to operate, since users have to navigate the palmtop's operating system and press keys or use pen-strokes to read their messages. These difficulties have to be balanced against expanded information obtained from the device. The ideal use for advanced pagers is the ability to broadcast information to many recipients at once. This is the notion behind Motorola's Embarc, which is not a standard paging service and requires some intelligent device to receive messages. Like a television broadcast, this is strictly for one-way, non-interactive applications to a wide audience. Any uses that require acknowledgment of this communications won't be appropriate or cumbersome for this market segment. Embarc differs from the one-to-one communications of less-capable pagers and beepers. Understanding this difference is critical to producing a successful application on these advanced devices. Second is the ease of nationwide roaming, or being able to receive data no matter where you are located in the US. Several paging service providers already offer nationwide roaming: SkyTel (Washington, DC) and MobileComm (Atlanta, GA) are two examples. Motorola's NewsCard offers nationwide roaming but configuring it isn't easy and requires some key (or pen) strokes. A third issue is the ease of integration into host platform applications. A very compelling business reason to use PC card pagers is the ability to manipulate the data received by the device in other applications that are running on the palmtop. For example, a message saying a meeting has been changed could update the calendars of all the participants automatically, as well as notifying them of the change. This is still far from what is currently offered in out of the box with any PC card device, although several third-party developers have begun work on various applications. For example, a Colorado bank service company offers its banking customers daily mortgage rate updates via the Embarc system: the messages are downloaded into custom databases and other applications that check to see if prospective home buyers can qualify for particular mortgages. 3. Nationwide two-way radio data network providers. Right now there are exactly two systems that are up and running: RAM Mobile Data (New York, NY) and Ardis, Inc. (Lincolnshire, IL). These companies have spent many millions of dollars to assemble radio towers and switchgear as well as billing systems and other infrastructure needed to provide coverage. These network providers are plumbing: actual users will probably not be so brand-conscious and look towards individual applications to force their decision on which network to use. More on these applications in a moment. Ardis has the largest customer base, reported to be over 30,000 users. Most of these are IBM and Motorola field service technicians (Ardis was initially formed as a partnership between the two companies, and TK latest info on what is happening here.) that use portable devices manufactured by Motorola. RAM has several thousand customers. Wireless modems that work on one network won't work on the other, although Motorola has made some announcements about future products that will work on both. And applications are usually device- and networ-specific: typically, applications developers choose one network provider and one wireless modem for their applications. The networks charge varying rates for usage. RAM currently has three pricing tiers: ranging from $25 to $135 a month depending on the amount of packets moved over the network. Ardis has similar pricing. Many marketing dollars have been spent to differentiate these two systems, but they are essentially the same and compete for the same sorts of customers: white-collar workers who want to run applications without wires, and move about the countryside as well. RAM and Ardis are both looking to replace and augment wired modems in their usage patterns: they have gone after typical modem-friendly applications such as messaging, disptach, tracking and file transfer. Both networks reach primarily metropolitan areas around the country, and coverage inside buildings varies depending on the elevation and building exterior construction. Several other networks are being installed for data communications: most notable is the Cellular Digital Packet Data (CDPD) network. CDPD is a loose-knit consortium of several cellular phone providers, including McCaw's (Kirkland, WA) Cellular One affiliates and GTE. CDPD augments the switchgear and existing cellular infrastructure by providing a digital packet network for data. It has begun service in a few cities but isn't expected to cover the nation for several years, however. And "begun service" is somewhat generous: most of the customers are still experimenting with various pilot applications, and no commercial rate cards have actually been published yet, with McCaw making individual deals with customers over pricing. CDPD is looking to augment traditional LAN-based applications such as transaction processing and client/server applications. This is because at the heart of the CDPD network is the Internet Protocol (IP) that are used by millions of computers of every shape and size around the world. This could prove to be a challenge for phone companies, who are used to selling voice and connection-based applications such as fax, to understand and market. The three networks (RAM, Ardis, and CDPD) share the same major issue of getting users and applications to work with their systems. It isn't easy, and requires the following steps: First, you need some way to connect your host computer to their network. This is typically done with a X.25 link (in the case of RAM and Ardis) or a dedicated T-1 circuit and router for CDPD. Next, you need the radio modem on the user end that will link to their network: all your mobile users have to have them. Finally, you need to modify whatever applications these users will use so they will run over the wireless link. All of this is time-consuming, requires multiple highly-skilled people, and is complex to get working. RAM has begun to make things easier by offering a wireless modem pool similar to the kind of service offered by many of the cellular companies. This would eliminate much of the complexity needed to connect the host computers to their network. But any application deployment will still require a great deal of skill and effort to pull off. In addition to CDPD, there is alot of movement afoot for what is called Personal Communications Services (PCS) frequencies. Basically, the Federal Communications Commission last September agreed to open up for auction several frequencies, including some that are presently used by other microwave uses. (Any winner of the these auctions will have to pay to change these existing customers to another frequency.) However, each of the service areas is to be auctioned separately, making a difficult, costly, and time-consuming process for any single vendor to obtain a nationwide license for these frequencies. Supposedly, the auctions will begin later this year. [TK: check this] This is a confusing and rapidly changing arena, and one where a great deal of money will be spent chasing too few customers for the next several years. And besides PCS there are a number of other two-way network providers that are building up their systems. These vendors include Pinpoint (Dallas, TX), Nextel (Rutherford, NJ), Metricom (San Jose, CA), and Racotek (Minneapolis, MN). Many of these had their origins in what is called specialized mobile radios or SMRs, mainly taxi and other vehicular dispatch systems. Each uses different technologies, spectrum frequencies, and has different power requirements and coverage areas. However, all are contenders to providing nationwide access for roaming users. The issues for these vendors are coverage and funding, in addition to the challenge of getting applications up and running on their networks. Since many of them grew from acquiring individual metropolitan dispatch operators, they have to grow their networks quickly and provide solid coverage across the country. Nextel has the coverage advantage, having licenses in 45 metropolitan areas already. Both RAM and Ardis have already built their own networks and are operating them today. Funding is also an issue: while RAM, Ardis, and the various CDPD have deep pockets and big backers (RAM gets most of its funding from Bell South, for example), many of these vendors will have a hard time obtaining revenue while they build out their networks. Metricom may have the advantage here: Microsoft has purchased a large stake in the company, and the company is barely breaking even from its first set of products, a wireless electric meter that has been popular with many utilities. Its network covers part of the Bay Area in California, and has devices that are similar to the Mobidem in size, weight, and battery life. The key difference is that its base stations can be placed on utility poles (both RAM and Ardis like to site them on top of expenisve large towers or buildings) and that they consume very little power. Nevertheless, Metricom will have to find customers quickly to succeed. Finally, no mention of nationwide networks could be complete with discussing AT&T's involvement. AT&T (Parsippany, NJ), while not providing actual wireless networks, is involved and supportive in two areas: first is with the General Magic (Mountain View, CA) alliance by providing a service called PersonaLink, which will connect all the various devices running MagicCap, the new operating system developed by General Magic. Products such as Motorola's Envoy when released later this summer will use Ardis' networks and built-in radio modems to talk to this service. And AT&T's Easylink makes use of RAM's network transports so that individual Windows users can connect to their electronic mailboxes wirelessly. Obviously, AT&T is playing the field and trying out various technologies. Easylink has never been widely popular outside of AT&T's own employees: both Compuserve and MCI Mail have more mailboxes and market share. However, Easylink was the first of the third-party email service providers to offer a wireless option. Motorola's Magic Cap device, called Envoy, isn't yet shipping, but we are gratified to finally see a device that has built-in communications as part and parcel of the basic package. While expensive (initial prices are expected around $1500), they might become a mass-market consumer item in a few years when the price drops and people understand what they are used for. 4. Two-way data-only radio modems for nationwide usage. These are typically used by roaming individual users and can be either data-only modems such as Ericsson GE Mobile Communications' (Totowa, NJ) Mobidem or Motorola's Infotac. These devices are alternatives to analog cellular modems and have several differences: First off, they operate completely digitally, unlike modems that covert data streams from digital to analog signals. Second, they operate over the packet radio networks maintained by RAM and Ardis, unlike the modems which use connection-oriented means. For example, if I use a cellular modem to send data to a host computer, I am charged for the entire time that I am connected, whether I send any data or not. With the Mobidem and Infotac (and the CDPD modems mentioned below), I am billed for only the actual packets that are sent, regardless of whether I am connected for minutes or hours to a particular host. Finally, they are dedicated for data, unlike cellular modems which piggyback on top of voice phones. These devices cost upwards of $700 and weigh about a pound. Battery life is about 5 to 10 hours, depending on usage, and are rechargeable and replaceable with specialized ones available from each vendor. A variety of software applications make use of these devices which we will get to in a moment. CDPD has its own set of radio modems, and of course these won't work on either RAM's or Ardis' networks. The first commercial product is a modem from Pacific Communication Sciences, Inc. (San Diego, CA) called the Ubiquity 1000 that fits inside an IBM ThinkPad 750. It weighs about half a pound and replaces the floppy drive of the laptop. It can run in four different modes: as a CDPD wireless modem, as an analog cellular modem, as a wired modem, or as a voice phone (either over cellular or wired connections) with an optional headset. All this functionality comes at a hefty price of $1495, which puts it out of reach of most business users for the time being. Don't have a Thinkpad? PCSI has a second CDPD product that is an external modem with the same functionality that will attach to any DOS computer. This is the Ubiquity 2000 and priced at $1595. Both have about an hour and a half transmission time on their rechargeable batteries, which according to experts could last several days of actual use -- but we haven't tested this claim. Cincinnati Microwave (Cincinnati, OH) has also announced CDPD modems, but these will be available through individual cellular operators. The MC-DART 100 is a CDPD-only modem and will sell for about $500. This device weighs less than a pound and requires a continuous power source, such as from a wall socket or a car lighter. Two critical issues here are coverage and roaming. The radio device is tied to a particular network, and if that network hasn't yet built any towers in your neighborhood you are out of luck. Coverage on all of these devices is still less than that offered by the analog cellular network, which has had a ten-year jump on building its infrastructure. Roaming is also an issue here: while the Mobidem can freely roam around the country and transmit and receive data without any user intervention, the Infotac is still a bit clunky, with Motorola promising upgrades for roaming Infotacs this year. We have not been able to obtain a CDPD modem to test its roaming capabilities. Both Motorola and Ericsson have promised smaller PCMCIA card modems for later this year, and that size will make these devices more competitive and convenient. In the meantime, these radio modems will be difficult to justify for wide deployment. 5. LAN wire replacement devices. There are a variety of products which are useful in places where network wiring (Ethernet or Token Ring) can't be used or where it is expensive to cable to the desktop: for example, in college campuses that span multiple locations in a given city where it would be costly to purchase a dedicated circuit to connect the locations or on a factory floor where running cables is difficult. Here the actual users are at fixed locations, unlike other wireless applications where users move about. The idea is that you replace the network cable and interface card with a special device that has a radio transceiver. This talks to a base station which has both another radio and a bridge back to the wired LAN. They have relatively short ranges, typically from 10 meters to 100 meters, with some products offering more. Examples of products include AT&T's (Dayton, OH) Wavelan, Motorola's (Schaumburg, IL) Altair, Solectek (San Diego, CA) AirLAN, Windata's (Northboro, MA) AirPort and National Semiconductor's (Santa Clara, CA) AirShare. There are a variety of frequencies and techniques used, from infrared to spread-spectrum to near-microwave. Again, the individual differences aren't as important as the overall functionality that these products offer: the ability to connect via ordinary LAN protocols users out of the reach of LAN cabling. For example, say that your office occupies two buildings that are separated by a few blocks in a downtown location. You could connect the two via a dedicated line that you lease from your local telephony company at several thousand dollars a month. Or you could connect the LAN in one building to the other via a radio link. You buy two of these devices and connect them to your wired LAN, either in a server or a workstation that functions as a bridge. Another example is in ad-hoc networking, where you want to drop ship a server and workstations to a remote site and have them up and running without having to worry about the cable installation. The market opportunities for these products are few and most have sold only a few thousand units overall. However, market researchers Venture Development Corp. (Natik, MA) claim this was a $200 million market last year, but I wonder how creative their accounting was. First off, most large corporations won't need very many of these devices, even if they have hundreds of computers that are already wired. Second, these products are expensive when compared to ordinary LAN adapters (costing several thousand dollars compared to several hundred, respectively). Even if you have some way of justifying the cost savings from cabling (and recabling) your office, they still are somewhat pricey. These products make the most sense for rapidly-growing businesses that don't want to invest in millions of dollars of cable plant in buildings that they will eventually outgrow, leaving all that wire in the walls. There are other issues for these wireless LANs: First and foremost is throughput. Wireless networks never obtain the throughput of a wired Ethernet or Token Ring segment, although they can come close to megabit speeds, depending on the technologies used and the physical obstacles in-between the transceivers. In our tests of the Altair network, we were unable to tell the difference between wired and wireless operations, however, so sometimes throughput can be made too important. Second is the support for particular protocols and network operating systems. While most of the products are indifferent to what network traffic they carry, some are more particular. For example, Wavelan does not support NetBIOS protocols, while Altair will run any Ethernet protocol. The critical information to examine is the number of different server operating systems drivers supported by each technology, since that is usually where these products are installed. Third is security from intruders. Most wireless LANs can be compromised merely by having someone nearby with similar equipment. For example, Proxim's RangeLAN (Mountain View, CA) operates on one of three different frequencies. Choose the right one and you might be able to snoop about. (There is an optional security ID that can be implemented to prevent snooping, but it isn't the default.) A few of the products have better security, but corporations worried about "wirelesstapping" might be better off spending the money on cabling. Finally, the issue of coverage and range is important. Each product has different ranges and coverage profiles, and actual mileage (more like footage, since these products don't trasmit very far) will vary based on building construction and other obstacles. Any serious purchaser should investigate these issues carefully before buying. 6. Finally, we arrive at various wireless software providers. This category is the broadest, and covers point-to-point applications, such as Traveling Software's (Bothell, WA) Laplink Wireless and AirSoft's AirAccess (Palo Alto, CA) as well as a variety of electronic mail vendors that have enabled wireless transports. The three largest LAN e-mail vendors (Microsoft, Word Perfect/Novell, and Lotus) all now have wireless client products that sell for about $200 per client, not including the cost of the radio modem and monthly network access charges. All three of these wireless email products run on RAM's network with the Intel Wireless Modem: this is a special version of Ericsson's Mobidem that is sold by Intel and uses a different set of protocols. None of these products has sold very well, mainly because it is difficult to provide the necessary wireless infrastructure and throughput is poor. The issues involved here are the kinds of client platforms available, the ease of switching between wired and wireless communications, and filtering and previewing one's mail. Let's look at each in some detail for both Microsoft Mail and cc:Mail. The attraction of these products is the ability to switch between wired and wireless modes of operation: For example, let's say I am in my office and connected via my LAN to my cc:Mail post office. I then disconnect or undock my laptop from the network and proceed to go on a trip. I'd like to receive some of my mail wirelessly when I am traveling, and then when I get to my hotel room at night be able to connect to my phone line and continue working. How easy is it to use these products? cc:Mail's Mobile client uses the same Windows (or Macintosh) client software and user interface that the non-mobile versions use. It takes just a few mouse clicks to switch configurations between wired and wireless configurations. However, the issue with cc:Mail is where your old mail is stored and how you will carry it with you. If you use lots of LAN email, you might not want to carry your entire mailbox around with you on your laptop: either for security reasons or because of the space all this mail will take up. Lotus has provided a docking feature in the Mobile software that allow you to move mail between the LAN's post office and the remote computer. This is somewhat cumbersome, but it does work -- you specify which messages you want to move when you come back to the office or just before you want to leave on a trip. Microsoft Mail is also different: you can store all of your mail on your individual workstation, unlike cc:Mail. This makes docking and disconnecting from the LAN effortless. Switching between wired and wireless modes with MS Mail is more cumbersome than with cc:Mail, however: you have to re-enter all your connection information each time you switch. Microsoft currently only offers wireless connections with its Windows client. Other issues for wireless email involves filtering and previewing messages. Since wireless connections take longer, you might not want to receive all your mail each time you connect. cc:Mail Mobile offers a preview feature, and you set the amount of time that a message header is displayed. During this time, you either accept or reject the message for subsequent downloading: if you do nothing, the message gets downloaded. With just a few messages, this can be helpful. For those email junkies that get lots of mail, this could be cumbersome. cc:Mail also has the ability to filter your messages. For example, you can set your mailbox to receive only messages that are smaller than a certain size. All others won't be received until you return back to your office or until you change your filter. This seems like a more useful function for those frequent emailers: it can run with little real-time attention while you are receiving your mail. Finally, Microsoft Mail does have a preview function, but you have to make two connections: first to obtain your message summaries and then to select the entire message to receive. Any filtering will require a separate add-on product that Microsoft OEM's from Beyond, Inc.(Cambridge, MA) and bundles with its Windows client. As you can see, going wireless is not simple nor straightforward. Both Microsoft and Lotus have just gotten started in this game, and the overall revenues they expect from the wireless side of the house will remain a small percentage for many years to come. This means that wireless email will always be an afterthought for these companies. This could be an issue for corporations that want to make wireless a mission-critical application. One place where wireless email is first and foremost is with Radiomail Corp. (San Mateo, CA), which has the widest installed base of client platforms, networks, and devices. Radiomail works on HP palmtops, Apple Macintoshes, DOS machines and even is found on Motorola's new Envoy. It runs either on Ericsson's Mobidem or Motorola's Infotac. The actual Radiomail software is a simple email client program: it talks to the back-end server at Radiomail's corporate offices. The advantages are that you can purchase all you need from one source and let them worry about the back-end connections: computers in Radiomail's offices are connected to the Internet via wired connections. Plus, Radiomail works rather simply: just turn on the radio modem and bring up your software. There are no phone numbers to dial, no passwords to remember. Your mail arrives quickly with little effort on your part. It is perfect for the executive on the go. If you already use email in your office, your Radiomail account will be a second mailbox. It costs less than $90 per month for unlimited messages, which compares favorably with other services. However, the company is small and has never been profitable. Currently Motorola is funding them and [TK: rumors are they will buy them soon]. Perhaps the best way to try out Radiomail is to purchase a Dell (Austin, TX) Latitude laptop computer: the DOS software comes bundled along with the first month's free subscription. Like Radiomail, there are a number of other small wireless software-only ventures. These companies have created a new breed of applications that will only work over wireless connections and have little or no comparison to any wired functionality. Products such as Air Soft's AirAccess is typical of this genre. AirAccess is a remote access program, whereby two DOS computers connected to radio modems (the software supports a variety including Mobidems and Infotacs) can synchronize and transfer files between them. The computers can both be mobile -- so that a work group could collaborate on a project without ever having to be in a fixed office, for example. And the two computers can be in different cities, provided they can access the wireless service provider. Laplink Wireless takes a different approach: here the radios operate only a few meters, so both computers have to be in the same room. However, the purpose of this product is to synchronize files between a laptop that is taken on the road and a fixed desktop machine. The cost of this product is $300, which includes both software and two radios manufactured by National Semiconductor. Summary. As you can see, the world of wireless data is changing rapidly and has lots of players. AT&T and Motorola will certainly be in the market for the long haul: wireless is mainstream business for both communications behemoths. The other and smaller companies will try to attract business, but it won't be easy: as you can see from our desrcriptions, there isn't any one-stop shopping experience and it is confusing sorting out what is needed and figuring out requirements for particular applications. BIO: David Strom is president of his own consulting firm in Port Washington, N.Y. He writes frequently for the computer trade press including columns for Infoworld, Communications Week, and Forbes ASAP magazines. He can be reached via the Internet at email@example.com.