The Wireless ATM Debate
Wireless ATM implementations have been discussed since wireline ATM equipment first appeared in 1994. The subject is surfacing increasingly as third-generation (3G) research and development projects move forward, and as broadband multipoint fixed services develop in scattered markets throughout North America. A debate is brewing on the future of wireless networks, and ATM has become an integral topic of discussion.
ATM in Use Today, ATM is used in private LANs, cable TV data networks, backbones for long-distance services and in digital-subscriber-line (DSL) systems for broadband residential and business services.
Now, it is beginning to appear in wireless networks as well, namely backhaul over point-to-point microwave linkages where it is a proven platform. It has seen limited use in satellite transmissions and is just beginning to appear in LMDS networks. ATM-based wireless LAN (WLAN) equipment and mobile products still are in experimental stages, and mobile networks are provisional and unproven. There are no industry-wide standards in place for over-the-air implementations in either fixed or mobile environments. Solutions that do exist are proprietary, and numerous critics say that ATM's validity still is questionable in the mobile environment.
But according to some industry observers, the marriage of the two is imminent because wireless networks already are tending toward both speed and diversity.
"The mobile networks of the future will have the speed to support the applications ATM is designed to facilitate," predicted Keith Shanks, Ericsson director of product management for D-AMPS.
This view still is debated in the cellular and PCS arena, but endorsed, at least provisionally, by Motorola and Nokia as well. According to Shanks, future voice and data networks will not be circuit-based. They will follow data protocols.
"We're looking at gradual increases in throughput speed starting at 384kb/s and extending ultimately to 2Mb/s, and we're looking at picocell architectures that will allow channel aggregation without taxing capacity," he said. "The mobile networks of the next decade will rival the capacity of LMDS."
Teligent already is using ATM in its 2GHz broadband multipoint networks, according to Robert Stuart, vice president of communications. The company will continue to use ATM in its infrastructure as it opens more markets and completes network build-out.
"ATM supports voice, text and video all on one network," he said. "It allows you to provision diverse applications."
Indeed, embedded ATM is fast becoming the standard among LMDS equipment manufacturers. ATM is the only way to manage delay in a wireless broadband network, according to Doug Smith, Nortel COO of broadband wireless access. This protocol has become a popular option among fixed-wireless service providers.
"ATM has already proved itself in point-to-point wireless applications," he said. "It's not that difficult to go to multipoint."
ATM also is appearing in WLANs and metropolitan area networks (MANs). For example, Phil Belanger, vice president of marketing, said Aeronet is working on ATM-based WLAN products including new 5GHz network equipment.
NEC also is developing ATM-based LAN and MAN products, and expects to begin selling them by year-end, according to D. Raychaudhuri, NEC assistant general manager of the computer and communications research division.
"The new 5GHz data networks will support voice, data and video streaming at rates up to 6Mb/s," he said.
He expects public carriers operating at these frequencies and offering broadband services to construct metropolitan networks. ATM will reside at the network core, and Internet protocol (IP) will ride on top of it.
Teledesic also endorses ATM. The company's 600 low-earth-orbit satellites will carry ultra-high-speed data traffic worldwide and will use an ATM-based platform, said chief scientist David Patterson. It is proprietary and uses double ATM cells, but supports the normal ATM virtual circuit and offers fiber-like performance.
"It's scaleable, flexible and robust, and it will best enable us to serve our customers," Patterson said. "It's the right platform for the future."
Multimedia ATM is pre-eminently a multimedia protocol, and ATM networks lend themselves to multimedia for several reasons. First, they establish priorities for delay-sensitive transmissions such as voice and video and do not permit text or graphics to interrupt voice and video, as happens in most packet-data networks. Second, they divide all messages into short blocks of equal length called cells, which create highly predictable traffic patterns. Finally, they negotiate transmission paths through the network before the message is sent and keep those paths open for the duration of the transmission, avoiding the "collisions" typical of most packet-data schemes.
ATM's theoretical advantages have been amply demonstrated in functioning commercial networks, and today, four years after ATM switching equipment was first made available, ATM has become ubiquitous in long-distance telephone trunk lines. ATM backbones now carry approximately half of all Internet traffic. On the carrier level, ATM has established itself and should continue to consolidate its position as the protocol of choice for high-capacity fiber pipes. But whether ATM will successfully extend its reach into last-mile applications remains doubtful.
"ATM presupposes multimedia, and I see that only in fixed-wireless applications," said Bill Frezza of Wireless Computing Associates. "Wireless ATM has been embraced by the 3G folks, but I don't see the market case. Where are the frequencies that will support picture phones on the beach and the kind of stuff they're touting?"
LMDS companies such as WinStar and Advanced Radio certainly support wireless ATM -- their infrastructures use ATM protocols -- but cellular and PCS carriers didn't offer clear support for wireless ATM.
"We definitely see the networks of the future going over to packet protocols for voice as well as data. But I'm doubtful that ATM will be used outside of fixed-wireless networks," predicted Ken Vaughn, AirTouch executive director of strategic technology. "I think IP will become the dominant protocol, and the fact that ATM has no presence at the desktop makes me doubt its ultimate survival."
Some hardware manufacturers echoed carrier reservations.
"We don't see a bright future for ATM over-the-air interface within mobile networks," said Haleh Motamedi, Qualcomm senior product manager of the wireless infrastructure division.
Although the company uses ATM in backhaul operations and will continue to use it there, the protocol's high overhead poses a problem in mobile networks.
"Eventually, they'll get fast enough to support it -- we see throughput rates exceeding 1.5Mb/s -- but we believe that IP, not ATM, will become the transport protocol for wireless networks," Motamedi said. "It's already part of our standard and the GSM data standard."
Heroic Measures ATM faces formidable obstacles in becoming the universal data protocol its creators intended. These include relatively poor efficiency compared to most of its competitors, lack of basic mechanisms for coping with heavy data loss a nd a general dependency on robust transmission channels.
"ATM over a point-to-point microwave connection is trivial," said Motorola's Lou Dellaverson, who chairs the Wireless Committee of the ATM Forum. "The protocol doesn't really know whether the connection is fiber or airwaves. Multi-point is a little harder, but the problems have essentially been solved. Mobile is what is difficult."
According to Dellaverson, that's because with mobile you're trying to provide guaranteed quality of service over a varying channel. How do you negotiate a virtual channel for a given service when you don't know that the channel won't fade? And what do you do when you hand off? How do you know you can negotiate a new channel that will meet the service requirements?
You don't, he said. And further complications arise once losses do occur.
"Data losses over wireless are ordinarily orders of magnitude worse than over wireline, so you've got to implement additional error correction beyond what's in the existing ATM standard," Dellaverson said. "Since you're dealing with multimedia, that error-correction strategy has to vary according to message. Voice can tolerate lots of errors but little delay, so you don't want to retransmit when errors occur there, while text can tolerate delay but not losses. The system has to be smart enough to figure out the appropriate strategy."
Seemingly, the high multipath resistance and soft handoffs associated with CDMA would provide a favorable environment for attempting ATM over the air, but Dellaverson suggested the combination would not be network efficient. Rather, ATM is best suited to TDMA schemes where cell length and time slots can coincide, and each cell can occupy a single time slot neatly. Perhaps for this reason, companies with a strong stake in the time division GSM standard, such as Ericsson and Nokia, have looked more favorably on 3G implementations of ATM. In any case, high data losses appear to be a given in wireless ATM mobile networks, and powerful error correction combined with dynamic negotiation schemes appear to be the only way to make mobile ATM work.
Despite these difficulties, Dellaverson insisted that solutions are imminent. NEC's Raychaudhuri agreed.
"Our 5GHz networks will support some mobility," Raychaudhuri said. "The equipment is in beta test right now, and is meeting our expectations."
But there is still an unanswered question: Considering the absence of carrier support, where will ATM wireless mobility solutions find employment?
Dellaverson isn't worried. There is a user group for wireless ATM within the ATM Forum, and there is demand on the enterprise level. He said military and public safety agencies also are interested.
Dellaverson also said that a wireless ATM facilitates completely distributed network management that would eliminate fixed-network access points and would allow the entire network to be mobile or essentially peer-to-peer.
"(You can) add infrastructure as instantly and cheaply as you need it," he said.
Perhaps so, but somewhere a carrier is going to have to make a commitment and to persuade customers to commit. So far that hasn't happened.
ATM is part of a family of fast-relay, data-transport protocols developed to facilitate high-speed multimedia services over public and private networks. Its cousins include frame relay, SMDS and IP version VI (the revised Internet protocol).
ATM is a collective effort of the wireline telephony industry to develop a data transport that would allow text, graphics and voice as well as high fidelity digital audio and video to co-exist and share network resources. ATM networks are digital networks by definition, and all messages -- whether voice, fax, text, graphics or video -- are digitized prior to transmission. However, ATM can co-exist with analog signals using the same physical infrastructure, and such hybrid applications occur in certain cable TV and DSL phone networks. ATM also can be used asa network core protocol for transmitting packet data. Because of the paucity of desktop implementations, observers believe that in the near term, wireless as well as wireline implementations of ATM generally will encapsulate native IP transmissions.
Want to use this article? Click here for options!
© 2014 Penton Media Inc.
From the Blog
Join the Discussion
Get more out of Connected Planet by visiting our related resources below:
Connected Planet highlights the next generation of service providers, as well as how their customers use services in new ways.Subscribe Now