Downloads |
The Internet has had a profound impact on the world over the last several decades. It has already touched almost every aspect of our lives and we are still in the early stages of the Internet revolution. Its open and innovative environment has seen the creation of such compelling applications as Skype, Facebook, Google Earth, YouTube, and Flickr to name just a few. The Web 2.0 phenomenon has truly been impressive as we've seen the power of collaboration, social networking, and user generated content. IPTV is now set to transform the entertainment business. The Internet has also spawned many innovative business models as well.
The next step in the evolution of the Internet will be to make it available anytime and anywhere. This will require the convergence of the mobile world and the Internet. This convergence is being driven by a host of powerful new mobile devices, high-speed mobile networks, compelling applications, and flat-rate all-you-can-eat billing plans. This confluence of events ─ this perfect storm - has caused mobile data traffic to grow at between 300 and 800% (depending on geography) in 2008. Recent forecasts by In-Stat have mobile data growing at 160% year-over-year for the next 4 years. Other forecasts point to even higher growth rates and almost all that traffic is headed for the Internet. Bottom line: the industry needs to plan for at least a hundredfold increase in mobile data traffic by 2013.
This tremendous wave of data traffic will require a completely new approach to building mobile networks. The industry will need to deliver orders-of-magnitude more traffic for an orders-of-magnitude lower cost per bit. The Cisco® IP Next Generation Network (IP NGN) strategy focuses on addressing the challenges associated with the rise of the Mobile Internet. Cisco has provided solutions to scale the wireline Internet, and we do the same for the Mobile Internet.
The Mobile Internet uses a rich tapestry of radio access technologies, including high-speed packet access (HSPA), evolution-data optimized (EV-DO), WiMAX, Wi-Fi, and soon the long term evolution (LTE), and IP binds these different technologies together. The good news here is that packet technology can scale in a very cost effective manner; the greater challenge will come in addressing the unique challenges associated with scaling the radio domain.
Network Infrastructure Transformation
Cisco is uniquely positioned to help operators scale their mobile networks. The move to all-IP end-to-end networks will certainly help increase the scale and lower the costs of the mobile cores. A focus on open systems with interoperability, long a staple of the IP world, will also reduce the cost of mobile networks. In addition, the closed and proprietary nature of cellular systems will slowly diminish.
A variety of approaches will enable the Radio Access Network (RAN) to scale. Some will involve advances in modulation, coding, and antenna technology such as LTE and WiMAX, whereas others will involve acquisition of new spectrum assets, but it is anticipated that the greatest success will come from local-area radio technologies such as femtocells and, of course, Wi-Fi.
So where do we start? Let's consider the critical components of a mobile network, and ask the question: Can they scale by at least hundredfold over the next 5 years?
• IP backbone: This component creates no problems because core routers such as the Cisco CSR-1 Carrier Routing System are designed to scale up to 92 Tbps, easily meeting the growth requirements of the Mobile Internet well into the next decade.
• Mobile gateways: This component introduces significant challenges because the rapid increase in data traffic means that these gateways need to be hosted on high-end edge routers.
• RAN backhaul: When voice was the dominant application, a couple of T-1 interfaces were enough. With the rise of the Mobile Internet, however, cell sites in major metropolitan areas will need anywhere from 100 to 200 Mbps in backhaul capacity.
• Radio domain: Scaling the RAN will be the most difficult challenge because capital expenditures (CapEx) for conventional cell sites are dominated by civil works, not something that responds well to Moore's Law. New macrocell technologies such as LTE will help, but they cannot approach a hundredfold increase in capacity.
The Mobile Gateway Challenge
Mobile gateways terminate mobile-specific protocols and connect users to the global Internet. A variety of different gateway technologies are used today, depending on the airlink. In the LTE world, the main elements of the evolved packet core (EPC) are the Serving Gateway (SGW), the mobility management entity (MME), and the Packet-Data-Network (PDN) gateway. The Serving Gateway provides local mobility, the MME supports the control plane, and the PDN gateway supports global roaming as well as providing the IP point of attachment to the mobile device. All three of these functions can be collocated in the same box, but that is a deployment matter. The Serving Gateway will most likely pushed out to the IP edge network to better offload traffic. The theory here is that not all traffic needs to be hauled all the way back to a tunnel termination point in the data center.
A significant challenge with LTE gateways revolves around performance. Because the circuit switching of voice is not defined for LTE, all voice will eventually need to be carried as voice-over-IP (VoIP) traffic. Next-generation modulation technologies and smart antennas will cause a lot more traffic to enter the gateways. The greatest challenge may well come with the rise of the Internet of "things" (aka machine-to-machine communications, which involves huge numbers of devices connecting to the Internet through a mobile network). The number of "things" could easily hit 100 billion or more in the next decade. "Things" include utility meters, tracking devices, sensor networks, appliances, consumer electronics, automobiles, etc. A very high session-activation rate will be required for their support, translating into the need for a very high-capacity gateway running on a highly available platform ─ an area in which Cisco has traditionally excelled.
What do we know about capacity requirements in the Mobile Internet? Some of the more forward-looking mobile operators are already dimensioning their networks for 30 kbps per provisioned user. This level may seem high, but it assumes large numbers of laptop users ─ and laptops can generate a great deal of traffic. Next up are a new class of devices called mobile computers of which the iPhone is the first instantiation. We are already seeing 3G iPhones drive 500 MB/month in usage, or a ten-fold increase compared to legacy devices. This rapidly increasing traffic load shows no sign of slowing down. To stay ahead of this trend, many of the more aggressive operators have begun to deploy Cisco mobile gateways running on the Service and Application Module for IP (SAMI) inside a Cisco 7600 Series Router. The Cisco 7613 Router loaded with eight SAMI modules can provide 75 Gbps in throughput, making it the only shipping LTE-ready gateway in the industry (see figure 1).
Cisco will be building Serving and PDN gateways to support LTE radio deployments. Both gateways will run on the SAMI module inside a Cisco 7600 Series Router, the most popular edge router in the industry with more than 70,000 deployed worldwide. This platform already hosts the Cisco Enhanced Gateway GPRS Support Node (eGGSN), Cisco Packet Data Serving Node (PDSN), Cisco Home Agent, Cisco Content Services Gateway (CSG2), and Cisco Broadband Wireless Gateway (BWG), and a fairly straightforward software upgrade will add Serving and PDN gateway functions. Cisco supports operators that are migrating from both the Third-Generation Partnership Project (3GPP) and 3GPP2 ecosystems, and our products will interoperate with LTE radios from the leading RAN vendors.
Our combination of IP expertise, standards compliance, performance, and platform flexibility makes Cisco a compelling choice for mobile operators, and scaling mobile gateway capacity a hundredfold will not be a problem.
Figure 1. The Cisco 7613 w/SAMI - The Only Shipping LTE-Ready Platform in the Industry
The RAN Backhaul Challenge
With improvements in airlink modulation, coding, and antenna technology along with additional spectrum we are seeing backhaul requirements start to increase very rapidly. As we look toward LTE, this number can easily hit 100-200 Mbps per cell site (see figure 2). Ethernet-based access networks are a cost-effective backhaul solution for all-IP networks. They also offer the advantage of being able to carry legacy traffic (TDM and ATM) via pseudo-wire technology. Pseudo-wire tunnels are built using IP/MPLS L2VPNs. Pseudo-wires are also used to carry Ethernet traffic in Carrier Ethernet deployments. This IP/MPLS layer can run over the top of fiber, microwave, or even T-1 based transport.
Whilst the economics of all-IP-RAN transport promotes a simple Ethernet access layer, L2VPN encapsulation for legacy traffic requires IP/MPLS capabilities be distributed all the way out to the cell-site. Cisco developed the Cisco MWR 2941-DC Mobile Wireless Router specifically for this role. It offers industry-leading scale along with support for both T-1/E-2 and Gigabit Ethernet interfaces. When coupled with the rich set of functions required for legacy traffic support, the Cisco MWR 2941-DC will allow operators to accelerate the deployment of future proof IP-RAN technology.
Cell-site routers are only half the solution; a scalable aggregation platform is also required. Our aggregation platforms include the Cisco 7600 Series Router and the recently announced Cisco ASR 9000 Series Aggregation Services Routers. Both routers are ideally suited to aggregating hundreds of mobile cell sites as well as wireline traffic of various types.
The future of the mobile core rests with IP technology. By getting on the price-performance curve of the IP industry it will be possible for mobile operators to drive costs down even as they carry orders-of-magnitude more traffic. A hundredfold increase in capacity is achievable with IP-RAN backhaul networks. The real challenge will involve trying to emulate this success in the RF domain.
Figure 2. Advancements in Airlink Technology Will Require Advancements in Backhaul Technology
The RF Challenge
The part of the network that is the most difficult to scale is the RF domain. In an effort to get out in front of this challenge 3GPP has been driving the development of LTE RANs. LTE, also known as evolved UMTS terrestrial RAN (eUTRAN), is an Orthogonal Frequency Division Multiple Access (OFDMA)-based radio access technology that is highly optimized for packet traffic. Standardization should be complete in early 2009, with field trials later in 2009, early deployments in 2010, and more broad-based deployments in 2011 and 2012. LTE is very similar to WiMAX, the biggest difference being that the LTE ecosystem focuses primarily on "paired" spectrum and the WiMAX ecosystem focuses on "unpaired" spectrum.
The move to new radio access technologies such as LTE and WiMAX will help to scale the RF domain, but they are not enough. Scaling the macro-area radio network can only partially solve the problem. Best estimates are that new modulation techniques and more advanced antenna technology might produce a tenfold improvement in capacity. In order to scale the RF domain a hundredfold over the next 5 years, we need to look elsewhere; macrocells and more spectrum will not be enough.
One factor to our advantage is that more than 80 percent of all mobile traffic originates from users who are indoors. Why use expensive macro-area base stations to serve users who are indoors and most certainly not moving? Femtocells have emerged recently as a very compelling solution for indoor coverage (see figure 3).
The femtocell provides an indoor base station with the form factor of a Wi-Fi access point and at a very low cost (ideally less than $150). Femtocells work over a very limited area and support a very limited number of connections. Used primarily in the home, they can be deployed in other locations as well. They are self-installed and AC-powered, and they use the Internet for backhaul. The advantage of this approach for operators is that they do not have to pay for backhaul, power, site acquisition, or new macro-area cell deployments. It also allows them to scale RF network capacity through very high spectral reuse. Users benefit by having a dedicated base station in their home. The few remaining technical challenges are being addressed, and we should see deployments in 2009.
Cisco is very focused on this space, and is investing in IP.Access in the United Kingdom. We believe that the tremendously high spectral reuse that comes with femtocells can address much of the hundredfold increase in capacity that is projected for the industry.
Wi-Fi will also have a big role to play in the build-out of the Mobile Internet. This very popular and very inexpensive radio technology operates in the unlicensed bands (offering several hundred MHz of capacity). Wi-Fi access points are popular in homes, businesses, and anywhere that people congregate in large numbers. It is available on most laptops and most high-end mobile devices. As costs continue to decrease, Wi-Fi will start to appear on lower-end mobile devices. Technologies such as Wi-Fi are well-suited to such bandwidth-intensive activities as streaming YouTube videos to users who are not moving. The industry continues to work on improving power control and connection management, among other areas. Cisco has been a clear leader in this area for many years, and we will continue to innovate in all aspects of Wi-Fi technology.
It is clear that both femtocells and Wi-Fi have a big role to play in dramatically scaling the RF domain through extremely high spectral reuse and leveraging the power of IP technologies and the Internet.
Figure 3. Femtocells in the Home
The Architectural Challenge
The move to next generation radio technologies like LTE and WiMAX will flatten the mobile networks. This flattening will greatly simplify the architecture, reduce cost, decrease latency, and improve the user experience. Functions such as the radio network controller (RNC) will collapse into the access point (also known as a base station) and the Serving GPRS Support Node (SGSN) will evolve into an MME (see figure 4). The future network will look like a routed aggregation network connected to thousands of access points through an Ethernet access infrastructure.
IPv6 will also play a significant role in the Mobile Internet. IPv4 address exhaustion will really come to a head with the rise of the Mobile Internet. With billions of new users and potentially hundreds of billions of new "things" we will need to move to IPv6. A critical focus for Cisco is not just supporting IPv6 in our equipment, but developing strategies for deploying IPv6 in production networks.
Cisco expertise in all aspects of scaling data networks will be in great demand as the Mobile Internet build-out accelerates in the years to come.
Figure 4. The Flattening of Mobile Architectures
Conclusion
The Mobile Internet will see the convergence of the Internet and the mobile world, and Cisco solutions can address the tremendous scaling and cost challenges. These solutions take full advantage of the scalability of IP along with some very compelling indoor radio solutions.
Cisco focuses on standards compliance, interoperability, open systems, expertise in IP, and a strong worldwide footprint. We can help your organization traverse this next phase of Internet evolution.