Broadband Wireless: The End of Wires

Most convergence is really just consolidation: Why hire someone to build a phone system, boardroom execs ask, when you can just pile more work on the IT architect? Why build separate networks for voice and data when a single one can carry both? Why pay phone bills when calls can travel free over the Internet?

Sure, VoIP vendors talk about new applications such as videoconferencing and collaborative document editing, but those are mostly niches. For most organizations, convergence is more about cutting costs than adding new functionality.

But VoIP does offer one compelling feature: Mobility. Unlike a conventional phone line, it can go anywhere where there's Internet access. Employees can receive calls on an office number and access the enterprise PBX while working from home, a hotel room, or if using a softphone, a Wi-Fi hotspot. It's both cheaper to the employer and more convenient to the employee.

Unfortunately for IT, employees have already found an even more convenient mobility technology: the cell phone. While cellular plans aren't as cheap as landlines, let alone VoIP services such as Skype, their coverage beats any tethered technology. Add new features such as push-to-talk or voice messaging, then roll those into GUI-accessible voice mail or centrex, and it's easy to see why people increasingly treat their cell phones as their only phones. To these users, the enterprise PBX--IP or not--is irrelevant.

And voice is just the start. Broadband technologies such as WiMAX will let wireless service providers offer mobile data straight to end-user devices, bypassing firewalls, switches, and servers in the same way that cell phones bypass the PBX.Although wireless broadband is often seen as a competitor to cable, DSL, or fiber, radio waves don't know the difference between fixed and mobile endpoints or between the LAN and WAN, so they also bring carriers into direct competition with enterprise IT.

To fight back, IT needs to understand the new wireless technologies and embrace them where necessary. Mobile WiMAX is still years away, but services based on other systems are already dissolving the boundaries between fixed and mobile access in some cities. Voice over Wi-Fi (VoFi) is an option for bringing cell phone users back into the enterprise fold, but it has its limits.

Most importantly, the 3G industry has developed an architecture aimed at extending the cellular carriers' concept of roaming to fixed and enterprise networks. Known as the IP Multimedia Subsystem (IMS), it provides a standardized interface for users of one network to access services on another while physically connected to a third. Enterprises who embrace it will be able to treat carriers in the same way carriers treat each other--as equals.

TIME DIVISION

From one point of view, the distinction between wired and wireless access links is a historical accident and a temporary situation. America just happened to build an extensive copper telephony network a hundred years ago, so it's now fully depreciated and relatively cheap to use. Conversely, cellular networks are much newer, so the companies that build them need to charge more to recoup their construction and spectrum licensing costs.The division doesn't exist in countries without universal wireline telephony, or where cellular networks have been around for longer. At one extreme, African nations rolling out phone networks for the first time increasingly turn to wireless because it's cheaper to build a single base station than to dig trenches for wires. At the other extreme, Scandinavian countries have moved over almost entirely to cell phones because they offer mobility at almost no cost premium.

The same is now happening in the United States as cell phones become increasingly competitive with fixed lines--not just as an add-on, but as a replacement. The main argument for running VoFi is that in most organizations, at least half of all cell phone calls are made from within the enterprise premises, so building an internal cordless telephony network can mean savings.

However, this assumes that calls from cell phones actually cost more than from landlines, something that's not always true. As cellular networks become more established, costs fall. It also assumes that employees will willingly give up cell phones that work anywhere for cordless phones that are limited to a single building. When mobile broadband arrives, Wi-Fi data could seem just as limited, and Ethernet even more so.

WiMAX FACTS

The great promise of WiMAX is that the same networks can be used for fixed and mobile links. An operator can start by offering fixed wireless services, then add support for mobility as coverage improves. Customers who want mobility and only need data rates of a few Mbps can use a PC card in a laptop, while those who need higher speeds can attach a directional antenna to the outside of a building.

So far, this is all theoretical. In April, Intel and Fujitsu both unveiled radio chips for use in WiMAX equipment, along with a large number of customers. But that's only the first step. No actual end-user products have shipped yet, and even when they do they won't be of any use until services are actually up and running. Few in the industry expect that to happen for at least a year.

Furthermore, the shipping radio chips and announced products are all for fixed access only. They're too big to fit into a laptop, and the standard that they'll use for mobile handoffs, IEEE 802.16e, hasn't yet been finalized. While Intel's official roadmap still calls for 802.16e-compliant WiMAX chips to be built into Centrino laptops by the end of next year, that now seems overly optimistic.

Still, some service providers are building mobile networks based on modified 802.16 hardware. Because the client equipment is expensive, bulky, and proprietary, it's rarely used for direct connections to moving users. Instead, it acts as a relay for Wi-Fi hotspots within a moving vehicle. Commuter trains are the most popular for two reasons: They contain large numbers of laptop-toting passengers with expense accounts, and they travel along relatively straight and predictable routes, which makes it easier for the network to coordinate handoffs between base stations.For example, T-Mobile's U.K. subsidiary has placed 802.16 transceivers at one-mile intervals along the tracks that link London to Gatwick Airport and England's south coast, as well as within the trains themselves. Passengers pay about $9 for Wi-Fi access during the one-hour, 60-mile journey. The system can theoretically handle up to 6Mbps, but like Verizon's 3G network the data rate is limited by the transceivers' T1 and DSL lines, not the wireless links themselves.

BROAD BEAMS

There aren't any similar systems planned for the United States yet, partly because business travel by train simply isn't as popular. But the architecture does illustrate how WiMAX can work together with Wi-Fi. In particular, it demonstrates that converging fixed and mobile access doesn't necessarily converge the LAN with the WAN. Though users might prefer to have a direct link to the network, it isn't strictly necessary, and Wi-Fi's ubiquity means that indirect links can be more convenient.

Sharing a single uplink to the mobile network has another important benefit. It allows multiple users to share a single bill, instead of paying a separate subscription fee for each client device. In some areas, small businesses and enterprise branch offices are already taking advantage of this, thanks to non-WiMAX mobile broadband systems.

While WiMAX gets the hype, other rival technologies are already up and running. The most advanced are made by IPWireless and Flarion Technologies, two start-ups that have had a lot more success abroad than in the United States. Their systems have been deployed on a large scale in countries from Japan to Finland to New Zealand and proven very popular.From the end user's perspective, there's little difference between the two. Where service is available, customers can get Wi-Fi data rates at DSL costs, all accessed through a PC card NIC that's designed for laptops but can also be added to a router for connection sharing. In theory, both also provide the mobility of cell phones, though in practice that requires widespread coverage, which isn't available yet.

NARROW TARGETS

Most of the two vendors' customers are CLEC-style service providers, which are initially targeting business and residential districts to compete with DSL. They're not covering highways or railroads, so people in a moving car or train often find that they can't connect. IPWireless even promotes its system as "portable DSL" on the principle that most people use data networks while stationary, not moving around.

However, both still lack a major U.S. carrier customer. IPWireless is more likely to be successful with GSM-based carriers like Cingular because its technology is based on a data-only variant of UMTS, the 3G upgrade path for GSM. Flarion is more likely to be embraced by CDMA-based carriers like Verizon because its technology is designed to work inside the same narrow spectrum bands as CDMA. Nextel shut down a commercial trial of Flarion's system earlier this year when it was acquired by Sprint, but then announced another one using IPWireless' system.

The other problem for both systems is that they're proprietary in all but name. While IPWireless' system is technically an implementation of a standard--UMTS--it isn't one that anyone else supports. The IEEE has announced plans to standardize Flarion's technology as 802.20 (or Mobile-Fi), but progress has stalled due to a lack of interest from other vendors. WiMAX does have the backing of standards bodies and a large number of vendors, so it could still win out even though 802.16e isn't ready yet.However, both IPWireless and Flarion still have cards to play. In IPWireless' case, European governments have actually sold off spectrum specifically for its type of UMTS. Carriers paid billions of dollars for this during the bubble, but did nothing with it because they and the cell phone manufacturers were more concerned with regular, voice-centric UMTS. As for Flarion, it has the support of major vendors, including Siemens as an OEM customer and Cisco Systems as an investor.

IF YOU CAN'T BEAT THEM

Whichever wireless broadband system eventually wins out, carriers won't find enterprise-class data coverage as easy to provide as voice. The problem is capacity. Because the licensed spectrum that high-quality coverage requires is scarce and expensive, the carriers will need to build cells very close together. For coverage in or near buildings, this requires the cooperation of building owners.

Wi-Fi offers a cheap and simple way to add connectivity, thanks to mass-produced access points and unlicensed spectrum. However, mobile operators aren't keen to embrace it unless they also have a mechanism of billing users for access through Wi-Fi networks. Without it, they see VoFi as a competitor.

This is where IMS comes in. IMS was designed for roaming between 3G networks, but it's been extended to cover roaming onto any IP network, fixed or wireless. While this can help carriers bill users connecting through what would otherwise be a free hotspot, it also helps enterprises because it lets anyone be a carrier.By implementing IMS, an enterprise can begin to act like a mobile operator. When users leave the enterprise's premises, they roam onto a carrier's network, but the enterprise keeps control of most location and presence data. A very large enterprise with enough spare Wi-Fi capacity in an area with poor 3G coverage might even decide to allow two-way roaming, which could mean carriers paying phone bills to the enterprise, not the other way around.

TRIPLE PAY

The IMS core is composed of three servers: Two contain directories, and one acts as a central controller for the whole system (see figure on page 30). The directories, known as the Home Subscriber Server (HSS) and Media Resource Function (MRF), keep track of users and services, respectively. At the center is a Session Initiation Protocol (SIP) server, known as the Call Session Control Function (CSCF). These are described as functions to emphasize that they correspond to tasks, not necessarily to physical devices.

On a small network, all three can run on the same hardware, while a larger network might require an entire cluster or grid of SIP servers for the CSCF alone. The servers in this cluster can be further subdivided by function and include some capabilities not yet found in SIP itself. For example, standard SIP messages are encoded as human-readable ASCII text like HTTP, while extensions for location tracking use the even more verbose XML. When sent over a bandwidth-constrained GSM or CDMA network, these need to be compressed into binary.

The HSS is essentially an updated version of the databases that all cellular operators already use to route calls, mapping phone numbers to cells. Many see these databases as their most valuable asset because they control how a customer can be reached. The HSS is potentially even more valuable because it can also include virtual presence information, precise geographic coordinates, and information about routing specific types of messages.For example, a person sitting at a desk might decide to have all incoming cell phone calls routed to a fixed phone, and text messages routed to an e-mail account. In both cases, the carrier acts as a gatekeeper because only the HSS knows the person's location. Although the calls or messages don't run over the carrier's radio network, the carrier can still bill the customer as if they did.

This makes control of the HSS critical, from both a privacy and a cost perspective. However, the MRF is also important. It keeps track of the IP-based services available to users, allowing whoever runs the IMS core to customize each user's experience. For carriers, this mostly means linking up with content providers, from Disney to Playboy. For enterprises, it could mean ASPs or custom intranet applications.

A company whose sales representatives need access to Salesforce.com or Siebel CRM OnDemand might provide MRF interfaces to these, but few enterprises would choose to enable access to the pay-per-view porn movies and pay-per-play Java games that account for much of cellular operators' revenue. The point here isn't to censor or filter traffic, but to avoid being billed for premium content.

There are a few caveats to such a scenario, of course. IMS is technically very complex, and in its original form required both Multiprotocol Label Switching (MPLS) and IPv6. However, it's been made to work with IPv4, and IPv6 isn't the barrier it once was. Most vendors have supported IPv6 for years, and in June it received a powerful boost when the U.S. government mandated all federal agencies to switch to it within three years.

A more significant problem is that the IMS industry is still very carrier-focused. Though some vendors such as BridgePort Networks and Ulticom also sell to enterprises, most IMS software is aimed at mobile operators and comes with a carrier-class price tag. However, IMS has also attracted significant interest from the open-source community, meaning that lower cost or free implementations may soon be available.Voices And Viruses

Mobility isn't the only area in which VoIP is following wireless. Converged devices and services enable new types of malware, thanks to the integration between devices capable of running arbitrary code and those capable of making phone calls.

Most people correctly believe that cell phones are relatively free of viruses, spyware, and the other problems that plague PCs. However, this is mostly due to a combination of the phones' secure OSs (few run Windows) and limited computing capability. For all the talk of cybersecurity, a black hat can do a lot more damage with a phone than a PC--both to public safety and to the bottom line of whoever pays the phone bill.

DIAL 900 OR 911

Cell phone users in Europe already suffer from malware that dials premium-rate numbers, and one virus even tried to create a network of zombie cell phones for a DoS attack on the Japanese emergency services number. PC users got an early taste of the same problem back in the days of dial-up connections, when rogue dialers would reroute Internet connections from a customer's ISP to a number that cost several dollars per minute.Premium-rate dialers were little more than a nuisance to enterprise PCs that relied on a LAN for Internet connectivity, and virus writers have largely abandoned them in recent years as even homes shifted away from dial-up. Convergence means they'll be back with a vengeance--and they'll affect enterprises as well as consumers.

Most PBXs block 900 numbers, but attackers are adept at finding ways around that. The most popular is to use an international number, which also has the advantage of avoiding U.S. laws. Many cellular plans block international numbers too, but the value-added services enabled by IMS provide another avenue of attack. One popular scam in the U.K. subscribes people to a premium-rate texting service, then charges them about $3 a message to receive spam.

Because of the relative security and simplicity of cell phones, most of these attacks rely on social engineering, such as Trojans disguised as ringtones or an auto-attendant that plays a busy signal so that people think their calls haven't connected. But the power of a PC combined with the weaknesses in Windows mean that even savvy users may be vulnerable on a converged system.

Pros And Cons

Mobile replacement for fixed services

Pros: Convenient * Provides competition in access line market * Eliminates need for two ISP subscriptions (fixed and mobile) * May allow gradual upgrade path (fixed to portable to mobile)

Cons: Availability depends on service providers and governments * Spectrum requirement means base stations must be close together * IEEE 802.16e and 802.20 standards don't exist yet

IP Multimedia Subsystem (IMS)Pros: Provides network-independent access to IP-based services * Roaming is possible between service provider and enterprise networks * Enables genuine unified messaging with a single phone number

Cons: Service providers can bill for calls made through enterprise IT departments * Requires infrastructure upgrades, including support for IPv6 and MPLS * Not yet supported by most service providers

Respond to Chief Technology Editor Andy Dornan at http://blog.networkmagazine.com.