Reaching Third Base
by Fred Dawson
Data applications are set to be the driving force behind third generation CDMA technology. Today base stations are at the heart of solutions being developed by vendors to marry new high-speed services to voice infrastructures already in place. Fred Dawson
Rapid advances in base station technology have made the choice of network suppliers a much more complex proposition. On the other hand, these advances look set to result in significantly more choice in terms of migration to advanced services.
Where, at the outset of the CDMA revolution, strategists were primarily concerned with ensuring that base transceiver stations and base station controllers met the mobile voice performance specifications of the IS-95 standard, today they are faced with choosing systems that take full advantage of ongoing gains in data and other advanced service features in fixed as well as mobile forms. Moreover, they have to be mindful of the fact that, no matter which path they choose, the shift to this focus on data will have profound implications for the long-term configuration of their networks.
"I think the fun is just beginning," says Irwin Jacobs, chairman and CEO of Qualcomm, Inc. "We'll be providing a much higher set of data rates, a whole new set of applications, higher efficiency of the system." With such changes come major alterations in network design, he says, adding: "I think we'll see a whole new set of distributed architectures where the base stations go directly onto a fiber ring."
Such thinking is common to new buildouts such as the $650 million nationwide CDMA PCS project being rolled out by Pegaso PCS in Mexico. Jim Edson is regional vice president of infrastructure operations for Latin America at Qualcomm's Wireless Infrastructure Division, which is the primary supplier to the project. According to Edson, the new base station technology used in combination with advanced fiber architecture provides a cost-effective approach through the startup phase and a long-term platform for both mobile and fixed wireless services.
Seen in the light of such architectural trends, data in the PCS industry has less to do with the long-running debate over what the applications are and how much people are willing to pay for them, than it does with the implications for the cost of network infrastructure, claims Matthew Desch, president for wireless networks at Northern Telecom, Inc (Nortel). "Data applications are exciting, but the business case revolves more around cost than anything else," he says.
"IP [Internet Protocol] technology as a basis for the network is a very, very cost-effective technology," Desch says. This, he says, is because packet switching rather than circuit switching is not only much more cost-effective for data but also offers "great advantages for voice as well".
Such thinking is new to the wireless industry, where operators
in most areas have their hands full dealing with price competition
while they try to build out their networks. But it's now,
as new infrastructure is being built, that the long-term picture
must be understood if operators are to avoid paying potentially
devastating penalties in the future, Desch warns.
You won't get much disagreement on these points from other vendors. The wireless industry is barking up the wrong tree when it confines its thinking about the evolution to data to traditional telecommunications terms, says Merle Gilmore, president of the Communications Enterprises Unit at Motorola.
What carriers should focus on is "the transition away from circuit switch-dominated hierarchical architectures to a distributed IP packet-switched architecture", Gilmore says. "This will facilitate a single architecture capable of delivering the voice services and the data applications whether you're using GSM, CDMA or any of the new standards that are being proposed."
Lucent Technologies' prediction is that wireless will be the primary method of access worldwide and that IP will be the primary means of voice and data communications within the next five to ten years, says Eric Sumner Jr., the company's ventures vice president in the Switching and Access Solutions Group. "Interconnection of all network elements at the IP layer is the key," he says.
Some carriers are already moving ahead with packet data over CDMA, notwithstanding the relatively slow (14.4 kilobits per second) data rate and still unproven market demand. "We're getting out on the cutting edge because we see data as a big part of our future," says Sprint PCS spokesman Ed Maddox. "There's an education process we have to go through, for ourselves and our customer base."
While Sprint will start delivering data on a wide scale during the first quarter of 1999 at 14.4 kbps, it plans to move as quickly as possible to a platform that will deliver data to mobile users at 64 kbps and over fixed links at rates of hundreds of kilobits per second in the near term.
Maddox says Sprint's wireless data agenda dovetails with the company's local access strategy on the wireline side, dubbed Interactive On Demand Network (ION), where voice and data combined over IP provide support for seamless integration of a variety of services.
PCS operators employing CDMA technology have a big competitive advantage when it comes to allocating extra bandwidth to data, given the efficiencies they're enjoying on the circuit-switched voice side, notes Jeff Belk, vice president of subscriber marketing at Qualcomm.
The company's research shows that these carriers are currently using between two and four of their 1.25 MHz channels for mobile services in any given market, leaving in excess of 20 MHz of spectrum for use in new applications, including fixed voice and mobile and fixed data services. "There's plenty of bandwidth to work with within the existing spectrum allocations when it comes to putting 3G-type [third generation] applications into play," he says.
But, while the latest generation of base stations coming on line from the leading CDMA vendors all support IS-95 in the circuit-switched voice domain, they are taking disparate paths in the evolution to ever higher speeds on the packet-switched data side.
The emphasis within the CDMA vendor community a year ago was on completing the IS-95B specification, which would allow operators to implement 64 kbps data access over the 1.25 MHz channels of their existing infrastructures via software upgrades without having to change hardware. This is supported by Motorola.
"By providing operators [with] an upgrade to the IS-95B technology, we can supply them with some of the same high-speed data applications that 3G will offer, all through a simple software upgrade to their existing CDMA equipment," says Moe Grzelakowski, corporate vice president and general manager at Motorola CIG's Cellular Systems Group.
An alternative data transition option being adopted by Nortel is the CDMA2000 Basic system, which has supplanted IS-95B as a bridge to third generation applications because it offers support for four or five users at 64 kbps, or perhaps even 144 kbps, per 1.25 MHz carrier. Nortel plans to provide CDMA2000 Basic products some time over the next 18 months, but will require new application-specific integrated circuits (ASICs) in the base stations, representing "about 20 per cent of the infrastructure on the ground", not to mention support from terminal suppliers, says David Murashige, vice president for CDMA marketing and management at Nortel.
Lucent is taking still another approach by leaving the base station alone but requiring a change in radios, according to Cindy Christy, vice president of Lucent's AMPS and PCS business. This can be done because Lucent's new base transceiver stations come with ASICs that can be programmed to support higher-speed data throughput over multiple 1.25 MHz carriers, essentially allowing three 1.25 MHz streams to be fed through a 5 MHz radio, which is the channel capacity envisioned for full 3G systems.
"We're proving that third generation radios can indeed coexist with today's 2G systems, thereby preserving the investments our customers have made in cdmaOne," Christy says. "The technology will enable incumbent operators as well as new licensees to offer two tiers of service from a common network platform, depending on customer demand for bandwidth."
Qualcomm has come up with still another approach which, if successful, could radically alter thinking about third generation technology. Rather than relying on a "nailed-up" access rate such as 64 kbps for a set number of users per channel carrier, Qualcomm's new high data rate (HDR) technology exploits the bursty nature of packet communications, creating a contention-based access system. The trade-off is a decrease in per-user throughput from a peak of 2.4 megabits per second as ever more users come on the channel, resulting in typical access rates in the range of 1.25 to 1.5 mbps according to the vendor.
"It's like Ethernet but without collision, which means you can support a lot of users," Jacobs says. Unlike 3G systems, which are designed to support voice and data access over the same 5 MHz channel, the HDR system requires that one or more of the IS-95 1.25 MHz channels is allocated to data only.
This is a better way to maximize data efficiency, Jacobs asserts. "With data you can burst out packets at very high rates and pass around access [to time slots] to multiple users," he says. "You don't have to compromise data efficiency the way you do when you share the channel for voice and data."
Qualcomm demonstrated the new system at the recent PCS '98 convention in Orlando, using upgraded versions of its IS-95 base stations. The company used FPGA (field programmable gate array) circuit boards or "emulators" in lieu of ASICs, which it anticipates will be ready for support of HDR upgrades in field tests by mid-1999, with commercial rollouts by the end of the year.
First, however, Qualcomm must find partners to license the technology, including handset and other terminal suppliers, none of which have stepped forward publicly to offer support for HDR. "There are obvious complications with a proprietary new air interface specification when you don't have any sense of how much support there's going to be," Nortel's Murashige notes, adding that the company hasn't "closed any discussions on HDR with Qualcomm".
The options in the evolution to data may be varied but they won't stand in the way of operators deploying new base stations. The good news is that all the leading suppliers have come up with new base stations that address a wide range of buildout requirements and offer a high degree of flexibility.
For example, the new Nortel CDMA Metro Cell, as well as providing
a platform for upgrades to full 3G, makes use of fiber to
connect the digital enclosure to the radio enclosure at distances
of up to 200 meters. This cuts signal loss as well as installation
costs for various indoor and outdoor configurations, Murashige
says. Operators can deploy the high-capacity digital equipment
frame of the Metro Cell in conjunction with minicell or Metro
radios, which means they won't have to change out digital
frames in instances where they have to move from provisioning
for low-density to high-density traffic, he adds.
One of the first players to exploit this option is U.S. company Chase Telecommunications, Inc., a CDMA PCS operator based in Chattanooga, Tennessee. ChaseTel, with C-block (15 MHz) licenses covering 98 per cent of the population of Tennessee and areas of six surrounding states, is using Qualcomm's new QCore 22 base stations in conjunction with the Alcatel 1000 S12 switching center, says chairman and president Anthony Chase. "It's extremely important to have the technical freedom to select top-tier equipment suppliers," he says.
One thing that appears certain is that operators can anticipate continuing dramatic changes in economies of scale as a result of ongoing advances in the performance of ASICs and digital signal processors (DSPs). "We have tremendous room for improvement and are moving ahead quickly," says Jacques Issa, a product manager for (DSP) applications at Motorola Corp. "The next step is to move from .42 micron to .25 micron [semiconductor circuit] technology, which we already have in-house."
Motorola's Semiconductor Products Sector and Lucent's Microelectronics Group are building a joint design center for next-generation DSPs in Atlanta under a new cross-licensing agreement. Just how far the shared expertise has already taken them was evident in recent product introductions by both firms, where chip integration resulting from smaller form factors and lower power consumption has produced dramatic gains in performance per unit.
A case in point is Lucent's new enhanced variable rate codec, a second-generation 8-bit speech coding device for the CDMA market that integrates a DSP and voice codec on the same chip. This creates room within the existing confines of handsets and base stations for manufacturers to add DSPs and other devices supporting features such as voice recognition, noise reduction and acoustic echo cancellation, notes Suman Swarup, marketing organizer for the Wireless Business Unit at Lucent Microelectronics.
"Improvements in DSPs lead to reductions in the cost, size and power consumption of equipment," Swarup says. "If a single chip can do what two once did, then if you look at the call processing application the space you are currently allocating to that can support [the] handling [of] twice as many calls."
A measure of how quickly things are advancing is the fact that the company now can support six 13-bit enhanced frequency rate (EFR) codecs per DSP, Swarup says. These advances will be pushed even further, he notes, adding: "It's not out of the realm of expectation that we could achieve eight or ten EFRs by the end of the year."
With such capabilities set to be supplied to CDMA operators by year's end, infrastructure manufacturers are driving hard to get 3G systems into the marketplace before the standards conflicts are fully resolved, Swarup says. As a result, he adds, ease of programming in DSPs is a major focus of development activities as manufacturers seek to guarantee their customers that changes can be made to accommodate standards without having to replace newly installed gear.
"What's important is that we keep migrating the technology to provide the best possible solution whenever a customer needs one, while protecting those capital investments for the long term," Swarup says. "The 3G options are based on software-defined radios operating on separate platforms that can eventually be collapsed into one standard, if the chips are properly designed." Given the uncertainties surrounding the migratory path to third generation systems, such thinking may be a comfort to confused strategists who have to make deployment decisions now.