As 4G approaches, David Perkins identifies the many convergence challenges for next-generation wireless networking. The corporate vice-president and general manager of Freescale also outlines some potential solutions.
Consumers of wireless devices, applications and services are first and foremost looking for simplicity. They want simplicity and access to their information, data and entertainment in a safe, secure way – especially their confidential and private information. Enterprises want that as well; they want to know that their employees are accessing confidential data in a secure way.
Security is also important to content providers because they want to make sure that their digital content is appropriately protected and distributed. Carriers strive to deliver stable, reliable networks that can deliver the promises of multiple services, applications and rich media content to consumers, wherever they might be.
As consumers, we want this in a personalised way, whether we are in the house, the office, the car or just wandering around. We want wireless access in many different ways. I might not be interested in watching television on a train or a plane using a mobile device, but many children would be.
All consumers have different needs, but there is one consistent theme: we all want simplicity. I want access to my email at 30,000ft in real time with all the attachments opening seamlessly in a small screen, but we're not quite there yet.
4G is still a dirty word, especially in Europe. 4G is a really dirty word in other parts of the world where carriers want to recover their investment in 3G infrastructure. By 4G, I do not mean just another radio access network. I mean a network of networks that integrates seamlessly and allows people to move between different types of radio technologies in a seamless way, with consistent access to applications and technologies. And that is a pretty ambitious type of network.
It requires multiple changes in technology and network infrastructure, in handsets and software. A huge amount of innovation is going to be necessary to deliver on that kind of promise.
I believe that a 4G network will depend entirely on a full internet protocol (IP) wireless infrastructure. This will give us the ability to fully exploit the capabilities of the internet.
CHALLENGES TO THE 4G VISION
The first major challenge I see in realising the 4G vision is power consumption. This is critical because we are adding multiple processing and communication elements to drive higher levels of MIPS (throughput) in mobile devices. All of these elements will increase current drain.
Additional hardware acceleration technology is going to be required to manage power in this kind of environment. I see the emergence and use of OFDM-based technology as crucial to managing some of the process streams and power challenges in these kinds of applications and devices.
Parallel processing techniques that will enable widely deployed orthogonal frequency division multiplexing (OFDM)-based technologies are essential innovations. With these techniques we can manage current drain and improve battery drain for better battery life. No one wants to see their movie end three-quarters of the way through.
The second challenge is spectral efficiency. This is largely a matter of availability. We have a couple of choices. In order for more spectrum to be made available, we will either have to re-farm existing spectrum in 2G and analogue broadcast TV or open up higher-frequency bandwidths. Further improvements in spectral efficiency can be derived from the use of cognitive radio. Dramatic innovations will be required to deliver on that promise.
Even with these steps, the 4G radio access network will need to provide significantly better spectral efficiency, in the order of 10MIPS/Hz compared with only 1–2 MIPS/Hz available today from 3G systems.
COST AND REVENUE CONSIDERATIONS
A third significant challenge in realising a 4G vision is cost. This is not just about infrastructure, operating or handset cost, it is also about the cost of deploying services. There are a variety of challenges in this area that come along with the network topology required for a 4G system.
First of all, to deliver the spectral efficiency and coverage required, we will have to see a dramatic growth in the number of basestations. To support the kinds of services that consumers increasingly expect, we will need as much as three times more basestations to deliver a ten-fold increase in data rate.
One way to reduce basestation density is by applying advanced antenna techniques such as MIMO and space-time coding (STC). These techniques can improve spectral efficiency to reduce the number and growth rate of basestations. They can do this and still achieve the kind of coverage required to deliver the bandwidth necessary for the applications consumers want.
There are capital costs associated with growth in the number of basestations required to deliver coverage at high data rates. On the handset side, there are significant challenges in continuing to drive down the cost of integrating greater and greater processing capability in multimode RF technology.
From a carrier perspective, the affordability of managing, billing and distributing content over these networks to drive revenue to recover those higher operating costs is another challenge in realising a 4G vision.
Everybody is still wondering what the killer 3G application is, and it seems like voice is still pretty good. But as we get into 4G technologies, mobile media players, internet access, broadcast technology and other types of corporate aggregations will become more robust and will drive average revenue per user (ARPU) in the carrier space.
MINIATURISATION AND PROCESSING CHALLENGES
Miniaturisation challenges include power reduction, cost, size and product development cycle. Multimode technology in 4G means we have to be able to hand off the different types of radio access technologies in a seamless way. There are significant software, billing, carrier interoperability and enterprise carrier interoperability challenges. On the multimedia side, it is obvious that with rich digital media content come dramatic processing challenges for mobile devices.
A wide range of advanced technologies will be required to enable 4G. For bulk CMOS process technologies, we will continue to evolve and provide higher performance at lower power as the semiconductor industry migrates from 90nm to 65nm to 45nm to 32nm technology. Many of the technological risks are not completely resolved. But the path to resolution is fairly clear over the next five to ten years with respect to these evolving bulk CMOS technologies.
On the RF side, for RF CMOS, we continue to scale RF CMOS to enable high-speed high-resolution A/D converters, and we are developing isolation techniques in the industry and at Freescale for multiple radio operation and single-chip die integration.
Further advances in integration are happening at the packaging level, where we are integrating the RF module and system package module. Increasingly, we are integrating embedded passives as critical elements for manufacturers to integrate multiple radios into a single device.
ADVANCED ARCHITECTURES FOR 4G
In WCDMA, we have two variables at our disposal: coding and time. With OFDM-based 4G, we use frequency, space and time as variables to extract the data. With OFDM, information is separated into small sub-bands, and the information in each of these bands can be signal-processed independently in a parallel fashion.
We are moving as an industry to increasing parallel processing. The net result of this is that a high-speed, low-power drain OFDM engine architecture can be developed to support 4G data throughput requirements.
OFDM is typically discussed in the industry as being a WiMAX technology. One of the reasons we see WiMAX as an important technology is not so much because we believe in the base business model behind broadband wireless access, but because, from the perspective of next-generation cellular technologies, OFDM will be an important technology to support the kind of data rate that we all expect in cellular technologies.
We are looking at very high levels of integration to enable 4G evolution and convergence. At Freescale, we have driven some of this integration in the 3G world by combining the baseband processor, the application processor, the power amplifier, the power management solution and some other ICs into single-chip Freescale solutions we call the Mobile Xtreme Convergence Platform or MXC.
In a 4G model we envision integrating an MXC platform and an OFDM engine to be able to support 4G types of applications with extremely low pin counts, small die size and small power envelopes.
AN ECOSYSTEM OF BUSINESSES
4G is not going to be driven by a single entity or organisation. It will require a tremendous number of partnerships and a robust ecosystem so we can exploit the capabilities that are available to us in wireless technologies.
Given the sweeping changes in the world of technology, it is going to require multiple standards bodies, corporations and government entities to come together to drive standards-based interoperability and the opportunity to deliver 4G networks. Governments will have to manage the spectrum in different parts of the world, and this will have a dramatic impact on how we can exploit the capabilities available to us in wireless technologies.
To enable triple-play merged services delivered over wired and wireless networks, equipment providers are looking for affordability, simplicity, interoperability and reliability as requirements that come to them from their customers, the carriers.
As we move from circuit-switched networks to IP networks, some of the challenges include packet acceleration, traffic management, data integrity, security and quality of service, which all represent different challenges for us on the infrastructure side compared to the traditional network infrastructure.
This notion of high performance in a very constrained power envelope continues to be a significant challenge for the semiconductor industry, as well as for equipment vendors and carriers. Open standards continue to be an issue in the semiconductor industry, and resolving open standards and interconnectivity issues is an increasing challenge for all of us.
IP NETWORK SECURITY, QUALITY OF SERVICE AND TRAFFIC CONTROL
Traditional equipment vendors have historically operated at layers 1–3. Wireline internet access is increasingly being challenged to improve security. Security has multiple elements, much more than just moving encrypted traffic at faster and faster rates across the network. Security is also about denial of service attacks and digital rights management. These are all becoming carrier problems.
Improved security and quality of service require providers to be able to identify video packets and prioritise them so that viewers get an uninterrupted stream of video content if they are watching a movie or TV show on demand. These security / quality of service capabilities are going to be key elements of how we manage the network in both wireless and wireline infrastructure.
All this is creating new challenges as we see a migration in next-generation networking to support elements above layer 3 up to layer 7.
From the processor perspective, looking at the roles microprocessors and network processors have historically played in networking applications, the network processor has been exceptionally efficient at driving performance at layers 2 and 3, but suffers a significant drop-off in performance in layer 4–7 protocols. A general-purpose microprocessor does not deliver the kind of performance networks require at layer 2, and does little better at layers 4–7.
Today's communications processors use hardware acceleration techniques to achieve better performance in the lower levels and do slightly better in the higher levels. But there is still a significant content processing gap in terms of microprocessor, network processor and communications processor technologies.
Clearly one of the requirements from a semiconductor perspective is the ability to provide a solution that does not just forward headers and IP packets. We need to inspect those packets, connect those packets and carry out stream processing instead of packet processing.
In terms of basestation size and cost constraints, there is a trend towards more basestations covering smaller areas while managing multiple power output limits, frequencies and standards.
Customers want to reduce power amplifier (PA) cost. The PA cost of a basestation is the single largest expense. Every operator and every silicon vendor is under tremendous pressure to reduce these costs. Freescale knows this space very well, because we provide high-power RF amplifiers for this market. We have an extremely high market share, and we are under-pricing our competition, resulting in average selling price (ASP) reductions averaging 15%–25% a year.
The laws of physics in the semiconductor industry do not support those kinds of ASP reductions just on the basis of a straight-line path in process technology. So more and more innovation is going to be required to drive proliferation of basestations for 4G networks.
To summarise, consumers, both at the personal level and within the enterprise, are driving the wide range of requirements for emerging 4G infrastructure. Clearly the appetite for seamless mobility is increasing as customers get a taste of what could be.
Apple and ABC recently announced that consumers can now watch Desperate Housewives on their iPod. Developments such as this are whetting consumer appetites for more – and this drives hand-off issues in networks. Consumers do not care if they are at home, or in a car or on a train; they want continuous service everywhere.
Other challenges include the effects of power consumption and efficiency, spectral efficiency, and the cost of deploying not just infrastructure but the services on top of it.
And for handsets, the challenges are myriad: miniaturisation; multimode technology; the ability of a handset to support multiple different air interfaces in a seamless way; and multimedia challenges that drive an increasing amount of processing requirement and drive up power consumption, which in turn has to be managed effectively.
There are financial challenges in realising that at every node throughout the ecosystem of providers – whether it is base technology suppliers, component suppliers, software or services providers or carriers – significant investment is required to enable the next-generation 4G network.
On the infrastructure side, more intelligence is required in the network if it is to provide quality of service and a quality user experience in an adequately secure way, not just for consumers but also for the providers that make that content available to carriers. This is driving complex service provider requirements, including security, quality of service and traffic control.
Finally, power in infrastructure is one of the most poorly understood issues in the wireless infrastructure industry today, but one that is becoming a high priority for major operators.