Who Makes What: LTE Evolved Packet Cores

When Light Reading last took a Who Makes What view of Long Term Evolution (LTE) in mid-2009, it was very early days indeed for the technology -- and still about midnight for commercial LTE services, as there weren’t any. Now it’s different. The first (very few) real pay-for LTE services are up and running, a lot more are going to follow very soon, and many operator trials are in progress. Further, some heavyweight product development has occurred as major infrastructure vendors fill out their portfolios to be ready for LTE’s imminent mainstream role in mobile telecom. (See Who Makes What: LTE Equipment.)

It’s a specific part of that heavyweight development that is the focus of this new Who Makes What: the LTE Evolved Packet Core (EPC). One reason for this choice is that the EPC is one of the two big new pieces of the all-IP System Architecture Evolution (SAE) that LTE introduces (the other being the Evolved UMTS Terrestrial Radio Access Network -- E-UTRAN -- or more informally the LTE RAN). Essentially, the EPC is what sits in the middle and makes the entire LTE system work.

As Gabriel Brown notes in his recent Heavy Reading report "Evolved Packet Core for LTE: Market Forecast & Competitive Analysis":

But there is also a wider reason. The EPC will be one of the first global examples of what the telecom industry has been preaching for years: an IP-based mass-market converged multiservice telecom infrastructure. It isn’t eventually just going to be about LTE or even wireless mobile, although the EPC has been developed to provide a core network for all wireless access mechanisms, including LTE, 2G, 3G, and non-3GPP. Longer term, the technologies and capabilities exemplified by the EPC -- essentially, high network intelligence centered on IP -- will migrate to the fixed and wireline world, too.

The EPC is a complex piece of systems engineering and naturally involves various subsystems and functional components. However, network operators will tend to buy their EPCs as whole entities (or at least as substantial part entities), either separately from the LTE RAN or with it under a single-source contract, so this Who Makes What concentrates on vendors offering complete EPCs, although it does note in passing some of the vendors supplying subsystems -- notably the Policy and Charging Rules Function (PCRF), as policy control in IP networks has sustained the rise of a fair number of specialists, and these have obviously extended their reach to embrace LTE.

However, this concentration on the systems level doesn’t mean that Light Reading intends to limit its coverage of LTE technology to this area only. Editors will be looking at many other aspects of LTE -- hardware and software, devices and subsystems -- in future.

As with previous articles in the Who Makes What series, the aim is to provide a reasonably complete list of major vendors in the particular technology, provide some technology background and to indicate what products the vendors have or are developing, and how they are approaching the market. Here’s a hyperlinked contents list:

— Tim Hills is a freelance telecommunications writer and journalist. He's a regular author of Light Reading reports.

Next Page: Becoming Real

Before delving into EPC technology and its vendors and their products, it’s worth putting some context around LTE itself. The most obvious point is that LTE has progressed from technical standard to first real deployments with almost unprecedented rapidity for a major infrastructure change. Only a couple of years ago the wireless industry was much preoccupied by whether LTE stood much chance of catching up with WiMax in the near future (see, for example, WiMax: What's Working Now); now the question seems quaint.

Thus the Global Mobile Suppliers Association (GSA) ’s August 2010 statistics indicated that 101 firm LTE network deployments are either planned or currently in progress in 41 countries, including three systems that have already been launched commercially (in Sweden, Norway, and Uzbekistan). Additionally, 31 operators are engaged in LTE pilots or trials, prior to official commitments, bringing the total number of operators investing in this technology to 132 across 56 countries. Around 50 LTE networks could be in operation by the end of 2012, and there is industry confidence that operators now believe that, as a technology, LTE is ready for prime time, and that there is a need for the capacity, QoS, policy, and other features it enables.

Light Reading’s research arm, Heavy Reading , expects the mobile packet switch core market, of which EPC is a part, to increase approximately 10 percent annually to reach $1.65 billion in 2010. Equally unsurprising is the way that the EPC has become a focus for major industry players, whether vendors or operators. A telling recent example is AT&T’s July 2010 announcement that it was increasing its roster of IP/MPLS/Ethernet/EPC Domain Suppliers to three big names: Alcatel-Lucent (NYSE: ALU), Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (NYSE: JNPR). (See AlcaLu, Juniper, Cisco Share AT&T Domain Status.) Announced in September 2009, AT&T Inc. (NYSE: T)’s Domain Supplier program is intended to make the relationship with its hardware and software suppliers more collaborative.

Market Drivers
The rapid rise in operators’ real interest in LTE as a 4G mobile technology may have various causes, and vendors may have different views, but most would probably agree with Gary Leonard, head of Mobile Marketing for Alcatel-Lucent’s IP activities, that mobile operators are struggling to support the tremendous growth and increased bandwidth requirements on their existing 2G/3G packet core infrastructures, made particularly acute by the rapid growth in smartphone and data usage. As a result, many are considering options to address the capacity bottlenecks in their networks, either by upgrading to a more current 3GPP release (such as HSPA/HSPA+ Release 7 or 8) or by deploying an LTE network.

“In existing GPRS/UMTS networks, the packet core serves primarily as a centralized traffic aggregation and transport function, with both the SGSN and GGSN supporting both the control and data planes,” says Leonard. “In an LTE EPC, significant architectural changes were made to enable a more distributed packet core and to separate the control and data planes into independent functional elements. So mobile operators are seeing the LTE EPC as a strategic and integral component in the way they offer new services and monetize their solutions. Moving to an all-IP network with specific and differentiated levels of subscriber and network policy, intelligence, and services creates new ways to add and extract value from the network. The network becomes integral to monetized services.”

So a key driver for LTE is monetization, because the EPC will give operators much more control over their networks and services, and the way that customers can use them -- and be charged for doing so. As Ziyang Xu, vice president, ZTE Corp. (Shenzhen: 000063; Hong Kong: 0763), points out that operators are increasingly faced with a mismatch between data-traffic growth and the revenues they currently derive from it, and that smartphones are making this problem worse (see Figure 1). Essentially, revenues and traffic levels are becoming increasingly decoupled and mobile telecom moves from being voice-centric to data-centric.

“So operators have invested a lot of money and resources in building and expanding their packet-switched networks to provide higher bandwidth and throughput, but their profit hasn’t been increased accordingly,” he says. “Mobile packet services have developed fast and have helped to offset lower voice-service revenues, but operators have still noticed a decreasing ROI on mobile packet services. EPCs, such as our ZEPS, can help to overcome this mismatch by building an intelligent and value-added pipe with a user behavior analysis system, for example.”

Next Page: Basic EPC Technology & Architecture

There is a very brief sketch in the earlier Who Makes What: LTE Equipment of the structure of the Evolved Packet Core (EPC) and how it fits into the overall LTE architecture, so this page takes that as a base and adds a few more items that will help to clarify what vendors are up to in product terms.

A point worth stressing is that, somewhat contrary to common telecoms parlance, although the EPC uses the word core, it isn’t referring to, say, the heavyweight underlying Layer 1 and 2 technologies and products that operators have spent so much money on in building their big core networks in the conventional sense. Instead, the EPC is essentially about the refinements of Layer 3 (IP) packet processing and control for a mobile broadband-application environment, and relating these to higher-layer aspects, such as service policy and charging. Obviously, however, the EPC does have implications for the capacity, performance, and so on of underlying lower-layer technologies and products, and vendors clearly see opportunities here for, say, product integration to optimize overall network and system performance.

The EPC might be viewed as a further stage in the continuing condensation of telecom networks around Layer 3. IP in origin was tightly bound to Layer 4 through such developments as TCP and UDP. Subsequently, MPLS has effectively extended the reach of the fundamental Layer 3 devices (routers) into aspects of Layer 2, and GMPLS has done the same into Layer 1. Now the EPC is effectively pushing the ability of Layer 3 devices into the higher service-oriented layers.

But, in the meantime, the basic EPC is just a software/hardware combination packaged in a box -- and these can be surprisingly small (see Figure 2), although real networks will require larger numbers of the basic physical units to provide the necessary scale as LTE subscriber numbers increase.

In functional terms, key elements of the EPC are:

Other important elements include Release 8 SGSN and the HSGW for CDMA evolution to EPC (eHRPD).

Note that the Service Architecture Evolution (SAE) Gateway refers to the combined Serving Gateway and Packet Data Network Gateway. These are all new elements: The SGW, PGW, and MME were introduced in 3GPP Release 8; the PCRF was introduced in 3GPP Release 7.

To expand these terms a little, using definitions provided largely by Alcatel-Lucent, but typical of industry usage:

A further crucial point about the EPC is that in all-IP in LTE all services are mediated by IP, from user terminal onwards. So the EPC handles the processing and switching of both voice and data (which embraces everything from email and text messaging to Web, audio, and video) as a single unified domain, in place of the use of two separate voice and data core domains as with 2G and 3G architectures such as GPRS, UMTS, and HSPA/HSPA+. IP was not conceived with this sort of universal applicability in mind, and consequently using it as the basis for massive networks providing high-speed real-time and media-rich services between mobile terminals is not a walkover.

So the EPC has many years of IP engineering development behind it to allow it to address the radically different and more stringent QoS requirements of 4G LTE compared to the cores of earlier mobile generations. This means providing capabilities such as nine classes of QoS and to offer a vehicle for delivering enhanced quality of experience (QoE) to end users, while dealing with millions of users and related data flows or sessions. The EPC will clearly put a lot of emphasis on product performance and scalability.

A final general point is that the EPC should improve network performance by separating the control and data planes, and flattening the IP architecture, thereby reducing the complexities of hierarchy between mobile data elements. This promise obviously has big ramifications for network OSS and service provider IT generally.

Standards
In general, vendors are happy with the current state of LTE standards. As Don McCullough, VP of Marketing for IP and Broadband, Ericsson AB (Nasdaq: ERIC), puts it:

“Our view is that the standards are just fine. They are done, they are sufficient to support implementation and commercial service. There is still some stuff to be worked on, and it is being worked on, and is probably a good source of job security for our standards people,” he says. “There will be interworking issues around radio terminals -- the early ones are only data and pretty soon there will be some voice terminals out there. Policy control is well-enough defined to get things going, but I think as services at these speeds and with these different applications roll out, you may want to increase the policy-control standards.”

Andy Capener, director of service provider marketing for Mobility Solutions, Cisco Systems, broadly agrees:

“Now that Release 9 is out, standards are pretty complete. I would say that the standards work still underway is mainly value-added on top of Release 9 -- such as Self Organizing Networks (SON) and things like that. Even the Voice-over-LTE capabilities are pretty much set, so we don’t see any standards roadblocks."

As always with a major new standard for technology, the industry has to demonstrate interoperability between different vendors’ implementations. Demonstrators are now in hand, with a recent example being the MultiService Forum (MSF) LTE Interoperability Event 2010, held at the Vodafone Group plc (NYSE: VOD) Test & Innovation Center in Dusseldorf, Germany, and a China Mobile Communications Corp. test site in Beijing. This included the interworking of EPCs provided by several vendors, including NEC Corp. (Tokyo: 6701) and ZTE.

And interoperability work is likely to occupy the LTE industry for some time, given the complexity of the specifications. It has been pointed out that a quick count of the specifications available on the 3GPP Website shows that the number of specifications that LTE nodes need to comply with is greater than 35 for devices, 56 for the eNodeB, and 41 for the EPC. These do not include IETF RFCs, 3GPP2 specifications for interworking with CDMA networks, and various other requirements from ITU and regional regulatory authorities. Hence, interoperability always remains a major concern, and is one of the prime areas being addressed by most operators in their ongoing trials.

And magnifying the task is that Release production continues unabated. Work on 3GPP Release 9 is currently in progress, and Release 10 (aka LTE-A and the full 4G specification) is scheduled for 2012. So a complex technology is being expected to mature very quickly. Apart from the challenge this creates for interoperability, it also makes it likely that operators may skip certain releases anyway, just as many operators chose to move directly from 3GPP R99 to R5 in their 3G deployments.

However, with a technology as complex as LTE, standards are never the entire story. As Alcatel-Lucent’s Gary Leonard points out, vendors differ in the experience and approaches they bring to IP and mobility, and this affects the way real products are designed and implemented within a standardized environment. In an EPC context, this affects how vendors are, for example, implementing the user and control planes, IP protocols, mobility protocols, IPsec, DPI, end-to-end network management, and policy control and management. This also affects how the RAN, backhaul, and core operate together (with end-to-end management), how much wireline capabilities can be leveraged for mobile and converged solutions, and how the next generation of all-IP high-performance high-bandwidth broadband networks -- mobile, fixed and converged -- are run.

Next Page: Vendors & EPC Products

Lists such as Tables 1 and 2 quickly show that the LTE EPC is attracting several different types of vendor.

Following the breakdown on the previous page of the EPC’s functional elements, this Who Makes What divides vendors into two classes: (a) those providing all elements of the EPC (system vendors) and (b) those providing one or more (but not all) of the S-GW, PDN-GW, MME, and PCRF (subsystem vendors).

EPC system vendors can be separated largely into three types: (a) the big full-range (including wireless/mobile) telecom infrastructure vendors (for example, Alcatel-Lucent, Ericsson, and, at the other end of the alphabet, ZTE), (b) the big IP-oriented infrastructure vendors (clearly Cisco Systems and Juniper Networks), and (c) specialists of various types and sizes (for example, Tecore Networks Inc. and Tellabs Inc. (Nasdaq: TLAB; Frankfurt: BTLA)). These different types of vendor inevitably are developing marketing pitches that play on their supposed strengths versus their competitors' supposed weaknesses.

These vendors supply all the main components or subsystems of the EPC as a complete product package. That is, the S-GW, PDN-GW, MME, and PCRF. Of course, they will supply a good deal more, such as the associated OSS and so on.

EPC subsystem vendors, of which Table 2 gives just a sample, are a more mixed bunch. Many are naturally specialists from the software, mobile, IMS, and OSS sides of the industry and focus on gateways, the MME, the PCRF, protocols, and so on.

There is also quite of lot of vendor activity in offering what could be called LTE-transition products and components -- such as gateways of various types (for example, for signaling) and the protocol stacks needed for interworking. Essential points here include providing (a) some basic functions that the all-IP EPCs will need (such as Diameter signaling), (b) aiding the interworking of the EPC with existing mobile or fixed networks, and (c) boosting IP-packet-handling capabilities, and so on. Examples of such vendors are IntelliNet Technologies and Traffix Systems. In the future, some of these vendors may offer more substantial EPC components -- IntelliNet Technologies, for example, has indicated that it intends to do so.

Table 1: Major EPC System Vendors

Industry hyperactivity

That part of the telecom industry surrounding the EPC has been pretty active recently, both in the pure industry terms of mergers, acquisitions, and the like, and in product output. On the pure industry side, there is very clearly a secular shakeout in progress as some vendors have disposed of mobile assets, and this naturally affects the EPC business; but a number of EPC-focused vendors have also fallen into the arms of the established infrastructure vendors. Some recent examples include:

Systems products galore

Almost all the major telecom and IP infrastructure vendors have made significant systems-side EPC-related product announcements over the last year or so, even if only by badge-engineering their acquisitions. Essentially, all the major infrastructure vendors now have a commercially available and up-to-date EPC portfolio. A summary list includes:

Subsystem innovations

Although EPC-related subsystems are not the main interest of this Who Makes What, the following points give some indication, through examples of recent product developments, of how suppliers to the major EPC systems vendors are innovating to meet the challenges of the technology. Typical concerns have been to boost performance (for example, scale and speed), ease migration from 2/3G, and add more functions and features in such areas as charging, QoS, and services supported:

It’s obvious from this list that the PCRF is receiving a lot of attention, as this is a key element in the provision and control of end services. For example, the specialist roaming company Starhome is currently implementing the roaming PCRF for the visited network (V-PCRF).

“We believe that this V-PCRF is significantly different from the home network PCRF (H-PCRF), as it related to roaming issues only, such as resolving conflicts between the home and the visited network,” says Shai Ophir, head of Starhome’s CTO Office. “The component will also need to manage the policies of multiple home networks, matching between the inbound roamer and the relevant policy rules.”

Next Page: Vendor Angles & Activities

The following pages present a very summary and high-level look at some of the activities and approaches of a number of the vendors listed in Tables 1 and 2. For convenience, the material for each vendor is roughly divided under the following very broad headings:

To avoid any misunderstanding, this does not provide a systematic comparison of the vendors concerned nor a detailed analysis of their product portfolios, and neither is there any intention to do so. That is the business of Heavy Reading, with which this feature has no connection. Instead, the aim is to give a flavor of each vendor’s thinking in order to give some insight into where they are coming from -- and hope to be going to.

Also, in any big topic like the LTE EPC, there is inevitably a lot of commonality between vendors on some matters, and therefore to save space and the tedium of excessive repetition, some common material has been removed. So the fact that vendor A mentions point X, while vendor B doesn’t, should not be taken to convey that vendor B necessarily does not agree with, or support, point X.

Themes and variations
Nevertheless, even with such a rough-and-ready approach, patterns do emerge. Using the classification of EPC system vendors given earlier as a basis:

it is very obvious that their EPC approaches are heavily influenced by their existing market positions. The full-range vendors with large existing installed bases of 2G and 3G RANs and core infrastructures are very concerned about ensuring a smooth eventual migration to, and interworking with, 4G LTE for their customers. Similarly, those with extensive installed bases of fixed broadband infrastructure are concerned with leveraging this service experience into a converged mobile environment based on LTE. And being able to offer a complete end-to-end LTE solution is bound to be a USP for at least some smaller operators.

In contrast, the IP-oriented infrastructure vendors see the EPC both as a new opportunity in its own right and also as part of the inevitable further shift in network architectures to all-IP, and therefore hopefully world domination, as they think that they do IP better than anyone else. This leads to approaches based on best-of-breed functionality, which certainly seems to be fostered by the tendency of major operators to split their LTE contracts into different functional areas.

The specialists also benefit from such splitting, as it also creates possibilities for exploiting synergies between product capabilities; for example, integrating EPC and backhaul capabilities into a single platform.

EPC marketers have some favorite buzzwords. The most common are:

EPC subsystems vendor
Before considering the complete EPC system, in the interests of giving a wider perspective, it is interesting to take a quick look at vendors with some key products within the EPC.

Tekelec, following its recent acquisition of Camiant and Blueslice Networks, has expanded its range of LTE offerings, which now run from software and protocol stacks to policy control and subscriber data management systems.

EPC philosophy. According to Randy Fuller, director of business planning, Tekelec views the LTE EPC as “the core of the core," but the SGW, PGW, MME, and PCRF will not be the only devices and functions.

“Ultimately, bits will be delivered over many different types of access networks, (2G, 3G, 4G, WiFi/unsecured broadband, and fixed broadband), and therefore will receive bits from many devices on those access networks,” he says. “Plus, there will be many adjunct specialty core devices -- DPI, Internet gateways, load balancers, and so on that perform specialized functions for certain subsets of traffic or subscribers. The bottom line is that data centers will have more than these four devices.”

Challenges and responses. Although the LTE market is still in an early stage, overall the company sees a trend for service providers to buy several products separately for a deployment. Examples are:

“We also see that vendors vary widely in terms of technology maturity and the percentage of the standards supported,” says Fuller. “For example, from a PCRF’s point of view, all vendors seem to be moving towards Gx, but almost every GGSN/PGWs implementation of Gx is different. The ideal is for everyone to first support the full standard as a baseline, and then add extensions for proprietary functions.”

Products and differentiators. Tekelec’s approach is to support much more than just LTE and its current trials, because it sees LTE as just one piece of the puzzle in network evolution. The company views itself as the only one with a portfolio solely focused on scaling the intelligence layer of all-IP networks, LTE or otherwise. Deployments can be made over GSM, CDMA, UMTS, HSPA, LTE, Docsis, and DSL/FTTx with a single product, for example.

EPC-related products include the Multimedia Policy Engine, the Diameter Signaling Router, and the Subscriber Data Management (SDM) product family. Product attributes include:

Next Page: EPC Systems Vendors: Alcatel-Lucent

Alcatel-Lucent

EPC philosophy. The company views the EPC as having a dual aspect: initially, bringing the concept of a new, high-performance, high-capacity all-IP core network to LTE, but also representing the cornerstone of a further evolution of this new mobile core toward a common all-IP core for all wireless (3GPP and non-3GPP) technologies, and also toward converged wireless and wireline environments, resulting in even greater efficiencies and cost benefits.

Gary Leonard, head of mobile marketing for Alcatel-Lucent’s IP activities, argues that the imminent arrival of 4G mobile, with its requirements for high data bandwidths and scalability to support hundreds of millions of users and billions of end-user devices, while ensuring premium QoS per-user, per-application, and per-data flow, is causing a fundamental rethink by operators and vendors.

“It boils down either to continuing to use existing -- essentially a 2G -- core technology for the evolution of mobile broadband, and hitting a brick wall at some point, or making a choice to use a different -- next-generation -- technology for 3G/3G+ renewal and LTE/4G introduction,” he says. “However, even the incumbent mobile-core vendors are realizing the necessity of changing their product architecture to address new sets of requirements. This is something we realized several years back and have invested a lot of R&D resources into addressing it.”

Overall, Alcatel-Lucent says that delivering a new LTE mobile core in the form of newly deployed, purpose-built EPC elements is essential to ensure superior network performance and quality of services and to minimize overall business costs. And such a forward-looking EPC is a cornerstone for further business evolution in mobile networks, allowing the creation of new business models and facilitating rapid deployment of new innovative services.

Challenges and responses. Leonard points out that, despite the huge interest in LTE and the large number of trials, operators do have alternative options to jumping into the technology. Given the uncertainty of the economy and the risks associated with deploying new technologies in a growing (but still immature) LTE ecosystem, some mobile operators are choosing to increase the capacity of their existing packet network by adding more core elements and upgrading their entire networks to a more current 3GPP release, such as HSPA/HSPA+ as per 3GPP Releases 7 or 8.

“When this upgrade path is chosen, mobile operators expect that any new packet core elements being deployed will also be fully upgradable to LTE,” he says. “Alternatively, there are other operators that are either trialing or planning to deploy an LTE network in the near future so they can increase capacity while delivering enhanced broadband packet data services. These LTE early adopters also want to ensure that the EPC network elements they deploy today provide the necessary packet core 2G/3G features and functionality so that the older packet core can eventually be replaced.”

For the EPC itself, a crucial task, now underway, is to prove the technology’s claims in real networks. This embraces such matters as:

Complicating this is the fact that operators -- particularly the major ones -- are generally splitting their LTE RFPs/RTTs into multiple and separate parts: There may be a RAN part, a backhaul part, a core part, an applications part, and a service part, with different vendors for some of them. However, as LTE spreads and smaller operators take it up, these are more likely to look at single end-to-end solutions.

Products and differentiators. In practical terms, the company regards the LTE EPC as a subfunction of its recently released Ultimate Wireless Packet Core (UWPC). This is scalable, all-IP, high-performance, converged 3GPP (2.5G/3G/3G+/LTE/4G) mobile packet core, and extends the company’s LTE/4G EPC by incorporating the gateway support node functions (SGSN and GGSN) to support simultaneously LTE and 2.5G/3G/3G+ (GPRS/EDGE, W-CDMA/HSPA/HSPA+) on a common converged core, which can be deployed in wireless converged environments (for operators who have heterogeneous wireless access networks) and also in wireline and converged wireless/wireleine environments.

Alcatel-Lucent is now promoting what it terms the High Leveraged Network (HLN) architecture (see Figure 3). This is intended to support the evolution of any type of network (wireless or wireline) toward a converged and optimized architecture in order to provide network operators with service convergence, resource optimization, and the capability to monetize the network. Key elements from this architecture are used in the company’s LTE EPC (and UWPC)

The UWPC Next-Generation Mobile Packet Core comprises several hardware platforms: (1) the 7750 Service Router-based mobile packet core gateways (for 3G, 3G+ and LTE/4G); (2) the 9471 Wireless Mobility Manager WMM, which performs the functions of the SGSN and the MME; (3) the 5780 Dynamic Services Controller (DSC) for the PCRF, dynamically managing network resources and QoS of individual subscriber services; (4) the 5620 Service Aware Manager (SAM), providing an integrated element, network, and service-aware management solution across the entire network, encompassing RAN, backhaul, and the packet core (see Figure 4).

The GRPS/UMTS GGSN and EPC gateway user-plane functions are delivered through a service-aware IP/MPLS routing platform, which provides advanced service data flow, QoS, and traffic management, 10 GigE line rate packet inspection and packet processing, and high scalability and capacity. LTE and 2G/3G control plane (MME and SGSN) functions, as well as network and subscriber policy control, are realized through an in-house optimized ATCAv2 platform with scalable, high-performance processing hardware and mobility management software. OAM and management planes are supported with a unified, end-to-end all-IP network management system, which covers the mobile packet core, underlying IP/MPLS transport network, and also the RAN.

An important aspect of the UWPC approach according to the company is that, by providing a single, converged mobile packet core that supports both LTE Evolved Packet Core (EPC) and GPRS/UMTS packet-core functions, issues of migration are simplified. Operators can continue with a renewed, scalable GPRS/UMTS core based on the UWPC elements to address existing capacity constraints, but with the flexibility to migrate to LTE later.

Alcatel-Lucent sees a number of differentiators in its EPC capabilities and approach. A general point is based on the view that the introduction of LTE will be somewhat analogous in terms of its network, service, and operator business impacts to the wireline world’s move from narrowband to broadband ATM and then IP access some years ago -- an area in which the company has been heavily involved. Toolsets, capabilities, and experiences gained in, for example, IP triple-play service delivery architectures, are likely to be useful as operators deploy LTE networks and services.

Leonard argues that a wide range of capabilities add up to significant differentiators for the company. These include:

“Transport is very interesting because you have to look at the capabilities of LTE as it relates to SLAs, QoS, and quality of experience (QoE), and how do you map an air-interface bearer to the mobile layer -- there is a whole LTE specification for a nine-levels QoS layer,” he says. “How do you map the air-interface (mobile) layer to the transport layer? Both Layer 3 -- and there are different technologies for that -- as well as Layer 2 are possible. We feel that we have a very strong story across all of those different elements, aspects, and technologies. It depends on the carrier, but we are seeing a lot of momentum and a lot of interest in a Layer 3 transport currently, but both layers work fine. The MEF work very well at Layer 2 or 3 and are being successfully and widely deployed.”

Next Page: EPC Systems Vendors: Cisco Systems

Cisco Systems

EPC philosophy. Cisco regards the EPC as being the next-generation packet core network for all mobile access mechanisms, including LTE, 2G, 3G, non-3GPP, WiFi, and femtocells -- and potentially even for wireline networks.

“For Cisco, we see it as a great opportunity. Our focus and expertise is on IP, all-IP networks. We can leverage our complete, comprehensive end-to-end architecture, not only in the evolved packet core, but in the unified RAN backhaul, the IP edge, the IP core, the cloud -- a complete all-IP solution,” says Andy Capener, director of service provider marketing for Mobility Solutions at Cisco Systems. “The Starent acquisition is just one piece of that, so Cisco brings a whole suite of capabilities to that story.”

On this view, the company’s lack of a RAN capability is largely irrelevant because that is essentially about access and connectivity -- necessary and very important, but not a future driver of profitability for operators, as that will be down to the IP next-generation mobile network itself. The huge growth of traffic from smartphones and other mobile devices makes the multimedia packet core network critical to providing a better user experience, for monetizing new services and applications, and reducing costs through optimization.

“The way that the mobile operators are going to increase service and revenue opportunities, deliver traffic, offer a common customer experience, and more efficiently drive traffic will come through that IP expertise,” Capener says. “We think the game is really coming into our home court. It is not just equipment that routes packets, it’s how you configure it, make it easy to use, minimize provisioning, provide billing, troubleshooting, and all those things. Having the expertise to do that at the IP level is going to be absolutely critical when it is IP end-to-end.”

Challenges and responses. Seamless migration from 3G to LTE/EPC on the same platform is seen as a crucial product requirement, and is required both by major operators with existing 3G networks and, increasingly, by the smaller operators worldwide that have yet to introduce 3G, or which are now doing so.

Performance is obviously a critical issue, not just in terms of raw packet throughput but also at the signaling layer in terms of the number of call connections that can be maintained, the rate at which they can be set up, and so on, as the EPC is very much a transaction environment. Products need to leverage both the data and the control planes to meet these performance challenges.

“And how to monetize and optimize this network as it grows? and leverage that core? That is where we talk about our network intelligence being able to both monetize and create services, and keep people from abusing a service -- and also to provide optimization to reduce the cost of the network,” says Capener.

Products and differentiators. The ASR 5000 mobile multimedia services platform (see Figure 5), acquired from Starent Networks, forms Cisco’s current EPC LTE offering. A key feature of the product architecture is that it eschews the common dedicated-blade approach, where, say, one blade type provides an MME function and another type provides an S-GW function, in favor of a distributed architecture. In this, there is effectively only a single, universal blade type, which can be configured by software to provide multiple functions as required -- either within the same blade or on different blades.

Cisco says that this gives a very flexible system, and operators can also turn on specific features (such as QoS and rate capping) as required and without affecting traffic flow or adding additional physical equipment. Scaling is done by adding additional blades, configured appropriately.

As an example of the flexibility permitted, it is possible to have an SGSN and an MME for 3G and 4G in the same element. With many operators looking at a best-of-breed model, but also with specific or unique needs, such flexibility fits well with Cisco’s IP focus. Additionally, it clearly helps to support 3G/4G migration, for example, and also to offer distributed or centralized approaches to the EPC itself -- basically, do you put all the EPC functions in one node (or a very small number) or distribute them among many nodes? This has become a big issue in network designing for LTE, and one that depends on a huge range of operator-specific factors, such as physical geography, numbers, or peering points, and so on.

This approach does require Cisco to produce its own hardware in a non-ATCA approach currently.

“The problem is that we haven’t found, for example, any off-the-shelf hardware that meets all the requirements of our customers. Obviously we would love not to have to build our own hardware, but we haven’t been able to find a substitute -- and the bar keeps getting raised. When those types of platforms do catch up, you have to do more because the market demands it,” he says.

Product development is oriented to enhancing a comprehensive IP architecture to provide all the attributes that an operator needs to support monetization and optimization of the traffic. Key attributes include:

Next Page: EPC Systems Vendors: Ericsson

Ericsson

EPC philosophy. Ericsson sees the EPC, with its signaling and subscriber control functions, as providing a way for mobile operators to consolidate their different types of access -- 2, 3, and now 4G -- into a single way of approaching their business.

“We see EPC as a critical part of the move from a voice and SMS world to a broadband-dominated mobile world. Its primary role is to help the migration through that upgrade path,” says Don McCullough, VP of marketing for IP and broadband at Ericsson. “We see two primary areas of that market. The first is CDMA operators that have to go to LTE quickly in order to compete with HSPA -- such as Verizon in the USA. The other big part of the market is the operator trailblazers that are determined to fight competition proactively and for which maybe HSPA isn’t enough -- such as TeliSonera in Sweden.”

But he points out that, even though currently the EPC is strongly associated with LTE and earlier generations of mobile technology, the company believes it will eventually have a very strong role in the convergence and evolution of wired access as well. In principle, it should not matter very much which Layer 1 technology is used to connect a broadband device.

“Just as smartphones today can go from WiFi to HSPA or LTE if you are in Stockholm, you ought to have that kind of movement with broadband devices in general. Obviously, some of them -- like a large TV -- aren’t going to get carried around very much. But we see that the access world should start to look very much like a single, rather than a multiple, entity,” says McCullough. “Our vision is that it is a converged approach, and eventually we will be able to handle both fixed and mobile traffic through there. Our view is that only Alcatel-Lucent shares that with us; everybody else in this area is focused primarily on mobile.”

Challenges and responses. With one commercial LTE service now operational, various supply contracts in place, and many trials underway, operator response to the current generation of EPC products and technology has been good, according to McCullough. Ericsson’s experience is that, in the initial phase of the LTE market, there has been a strong tendency to pair the RAN and EPC sides, either completely with a single vendor or partially between two. TeliSonera, for example, uses an Ericsson EPC and RANs from both Ericsson and Nokia Siemens Networks. Such pairing helps to avoid finger pointing between different EPC and RAN vendors if problems arise with what is still a relatively new technology.

“Obviously, we are fully committed to interoperability, and we do think that, as the market matures, it will move to an interoperable one, but operators seem to be making pragmatic decisions,” says McCullough. “They may put speed to market and simplicity ahead of interoperability, but interoperability is still a huge priority, and nobody is going to get away with putting together an EPC that only works with their products. And we certainly aren’t doing that.”

Products and differentiators. Nevertheless, Ericsson sees its established position as a major supplier of wireless infrastructure as an important differentiator and strategic advantage as it means it can supply a paired LTE RAN/EPC solution immediately, with lower risk and faster deployment, and also provide an upgrade and migration path to LTE from a large installed base of earlier-generation radio base stations and gateways, particularly GGSNs.

Supporting this is a view that considers the EPC to be a software product, to be run eventually on three platforms: the current SmartEdge and also two new platforms to be supplied by Juniper Networks. This software emphasis is crucial because it enables Ericsson to meet the migration needs of a diverse group of operators. The company claims over 40 percent of the world’s installed base of radio base stations, and many belong to operators that will not upgrade to LTE for a long time and are still looking at HSPA as their next major upgrade, for example. Others, in contrast, are moving very quickly to LTE deployments. A software approach allows upgrades to be flexible and avoids a rip-and-replace approach to the evolution of the company’s embedded product base.

Next Page: EPC Systems Vendors: Juniper Networks

Juniper Networks

EPC philosophy. Juniper views the EPC and its associated architecture as a step change in the evolution or transformation of mobile networks. In a broad sense, it moves the network much nearer to where the company positions itself, with its focus on MPLS and IP, and to the capabilities and solutions it offers.

“As the mobile network evolves, the things that are going to be important are scalability, predictable performance, and how you monetize the network to deliver new services quickly, as opposed to being reliant on voice and SMS, which are the main drivers of revenue today. Those are all the critical components of what operators are looking at and where the network is headed,” says Dilip Pillaipakkamnatt, director, product management, at Juniper Networks.

Further, longer term the EPC architecture is set to have as wide an influence in telecom as IP-based network integration or convergence spreads and moves both horizontally and vertically. In horizontal convergence, multiple network functions are combined into the same network node -- in an LTE context this would include things like integrating the RAN-facing MPLS-PE into the EPC gateway, or integrating NAT or firewall functions that face the Internet side into the EPC gateway. In addition, EPC functions will get distributed. The wireline world experienced something similar when it made the transition from narrowband to broadband access, and Juniper expects LTE networks to follow this pattern, although it will take some time to complete.

Vertical convergence involves both wireless and wireline networks. This has already occurred at the backbone level in many operators’ networks, and the next step -- a common aggregation or backhaul network for both wireless and wireline -- is also well underway.

“But there is the piece left in the middle -- the services core. That, we believe, is the next step, but it may take a while for it to happen,” says Pillaipakkamnatt. “Operators are starting to ask questions about it, we are starting to see some RFPs and RFIs with questions on integrating the service edges. It will happen, but may take a little longer than the horizontal integration, probably due to organizational considerations within the operators’ environment or because the mobile and wireline nodes scale differently, and other such technology considerations.”

Overall, the introduction of LTE in response to the increasing demands of data on mobile telecom is forcing network operators to rethink their entire end-to-end network architectures. One consequence is that operators are now stratifying their RFPs to separate the RAN from the EPC because of the different economics of the two domains -- the RAN, being geographically extensive, costs far more than the EPC, but its influence on revenues will be much less.

“We believe that 10 to 20 percent of what operators will spend on the IP infrastructure will drive 70 to 80 percent of the future revenue stream as they move from voice, SMS, and connectivity to launch new services. This is a very significant change that has happened for the first time. The RFPs are reflecting the decoupling, which is making the operators build best-of-breed capabilities,” says Kittur Nagesh, senior director, mobility solutions, at Juniper Networks.

He argues that this makes the EPC a real asset for the operator and that, when designed well, it will accommodate both centralized and distributed functions to provide real-time subscriber knowledge and real-time orchestration, together with such network capabilities as bandwidth, QoS, security, social networking, context, and location services.

“We believe that this is a gold mine for the operators, and therefore not all EPCs are the same,” he says. “It is not just about implementing 3GPP standards, it is -- How do you convert the EPC into an asset that you can make money from? How do you bring the OTT players into negotiations and partnerships to drive incremental revenues? And so on.”

{Image 6}

Juniper formalized its LTE ambitions by announcing its Project Falcon in early 2010, with the goal of taking advantage of its existing position in IP edge routing, the Junos SDK (which enables customers and partners to develop applications for its mobile gateways), and its suite of mobile control and broadband packet gateway software. It sees that a combined control plane (2G/3G SGSN and MME) and a combined data plane (2G/3G GGSN, PGW, and SGW) offer deployment flexibility for operators as well as an opportunity to optimize subscriber quality of experience across the range of 2G, 3G, and 4G radio access.

Juniper sees the SDK approach as an important part of its strategy, as the EPC gateway is a natural point in the network for developing and trying out new services, and the Junos operating system itself embraces end devices, network elements, and the network as a whole. By providing a single SDK with open APIs, the company hopes to offer operators both a highly scalable EPC platform and a supporting open ecosystem of service and application developers with which to exploit and monetize it.

Junos thus plays a central role in Juniper’s EPC strategy, and will also underpin the approach to LTE migration, as the company expects to be able to handle many of the issues that migration raises through software upgrades, rather than by system replacement. Additionally, Junos includes capabilities that are important in the LTE context, such as security -- the recently released Junos Pulse network client, for example, provides secure access to corporate networks and cloud-based data and applications from mobile devices and smartphones.

On the hardware side, the company emphasizes the three-dimensional scalability of its EPC edge routers and service gateways in terms of the per-node bandwidth, number of subscribers hosted, and number of services. Underpinning the scalability is its new Junos Trio chipset, released in late 2009, which combines microcode programmability with an ASIC structure. This, Juniper says, allows it to combine the speed of the ASIC approach with the flexibility and intelligence of software to produce a layered structure of silicon / Junos / SDK / APIs / Services.

According to Pillaipakkamnatt, delivering services at scale (both control and data plane) will be the company’s focus for product development. Juniper’s NPU-based approach is at the heart of its ability to provide scaling with services. This differentiates Juniper from CPU-based gateways as software features require CPU cycles, reducing performance. In contrast, NPU-based gateways not only address the scale issues but also guarantee the QoS and low latency required for next-generation services. Finally, Juniper’s product is well suited to distribution as the EPC functionality will be delivered across the MX Series product range.

Next Page: EPC Systems Vendors: Tellabs



Tellabs

EPC philosophy. Tellabs sees the EPC as providing a range of essential functions for the mobile Internet or data network, and these functions occur at different levels of network activity or operator interest. At a very basic level, the EPC acts as the gateway between the wider Internet infrastructure and the complexity generated by large numbers of multiple and multifarious mobile sessions that are continually changing as user devices move and power up and down.

But the EPC is also where you can find a lot of intelligence that can be used to exploit a wide range of other capabilities for network control, service offerings ,and the operator’s business overall -- such as traffic optimization, QoS differentiation and prioritization, service innovation, and the like.

“In a nutshell, I would say that it is a very central piece of the whole network, which we know is going to grow dramatically,” says Rehan Jalil, director of Tellabs’s Mobile Internet Business Unit. “Hence we are already focused on doing a lot of investment in providing value to our customers. How to build the network. What features to provide. What capacity to provide. How to make money. How to optimize the network. So we are very focused on R&D, and the whole company is behind making it happen.”

And the importance of the EPC within telecom can only grow because of the potential for the spread of broadband Internet connectivity within the already huge numbers of mobile users -- 4.7 billion in 2009 according to the ITU’s "The World in 2009: ICT facts and figures." Of these, only about 14 percent had any form of broadband capability.

“Imagine when all these devices are converted and Internet enabled. Mobile will become the Internet, and the boundaries between wired and wireless will start disappearing. And the packet core, where all the sessions are terminated, sits right in the middle and will play a much wider role in how you take the session across the boundaries for mobile and fixed. And what roles you start playing for the fixed,” says Jalil. “It becomes a lot more interesting. So it is a very strategic investment from the perspective of the company.”

Challenges and responses. Operators are still at early stage with their involvement with, and deployment of, LTE and the EPC itself. Issues surrounding the latter include technical performance, flexibility, and monetization.

Jalil argues that technical performance and flexibility are linked because no one yet knows how these networks will be used a few years down the line, and therefore what the performance demands on them will be -- think of the impact of the iPhone/iPad on data usage, for example. Although the data-plane performance thus receives a lot of emphasis, the control-plane and messaging performance are also critical.

“These smart devices are very chatty -- if you launch an application, say, on the iPhone, it may behave very differently from Web browsing. If it is some form of a tracker it may go to sleep, come back up, and keep doing that the whole day. That puts a very different load on the network,” he says. “So what we need are platforms that have multidimensional performance -- for data, messaging, content handling, scalability of connected devices, and so on. That is one thing that has come up from operators. They are now realizing that the user patterns are still going to be evolving -- and dramatically evolving.”

And this uncertainty leads directly into another key challenge for the operators -- how to make a financial return from, or monetize, the large network investments that are required. Tellabs’s view is that the LTE network will have to become smarter to allow operators to move beyond just supplying a mobile broadband bit-pipe -- and the EPC is a crucial enabler of this because it can help operators understand network usage, prioritize and control traffic, increase efficiency, and support service bundles and creation.

The company argues that this view of the EPC is supported by the fact that major operators already see the EPC as a potential strategic resource and thus are going for best-of-breed approaches for future deployments through splitting LTE RAN and EPC RFPs. In contrast, however, smaller operators may well opt for the organizational simplicity of a single-supplier approach for a combined RAN/EPC package.

Products and differentiators. Tellabs’s EPC products are the SmartCore 9160 and 9180 Platforms. These are essentially smaller and larger versions, both physically and in throughput, with, respectively, in input/output terms, up to 40 1Gbit/s ports or 4 x 10Gbit/s ports, or up to 100 1Gbit/s ports or 10 x 10Gbit/s ports. For both devices the different port rates may used in combination, with both copper and fiber options. The 9160 is billed by the company as the densest 4G packet core platform in the industry.

Tellabs does not produce an LTE RAN, and sees its focus solely on the LTE EPC as an important differentiator, as this enables the company to concentrate on tightly integrated hardware and software to meet the multiple performance requirements of the EPC within a dense equipment footprint.

“We have built something that is very custom designed for packet-core applications in terms of data and messaging performance, DPI, Layer 7, encryption, and IPSec security, for example,” says Jalil. “Unlike some other vendors, which will sell you four or five cards to do the same thing, we provide one blade that has all the dimensions of performance. We are not just trying to leverage our previous router or server.”

And as a major provider of mobile backhaul, the company sees considerable potential for further product differentiation and customer benefits by integrating backhaul and EPC-specific capabilities -- for example, by exploiting the network usage data that can be gleaned from the backhaul system, and vice versa.

This approach of EPC focus and integration also supports the monetization and optimization capabilities needed to meet key operator challenges already described -- by enabling smart networks that can handle a very dynamic network environment, with multiple types of users and services.

Next Page: EPC Systems Vendors: ZTE

ZTE

EPC philosophy. ZTE sees the LTE EPC as a crucial part of 4G networks, but it is not only for such networks. It brings operators a simpler and flatter all-IP network that reduces operational expenditure, and is the first network architecture that can support the multi-access by different radios, including the 3GPP-defined types and non-3GPP-defined types. Further, the EPC can support service continuity for users roaming between different radios and different core network domains, such as the circuit and packet domains; it also introduces new features such as a dual stack for IPv6. And, for operators considering constructing a multi-access mobile network and wanting service continuity between different radios, the EPC network can be introduced before LTE radio if need be.

Challenges and responses. The company argues that LTE/EPC deployment is important for operators wanting a leading position in mobile broadband or mobile Internet services, but beyond these operators the market drivers may be different.

“It isn’t too pressing for operators with mature HSPA/HSPA+ networks or in markets where mobile broadband is not widely demanded,” says ZTE vice president Ziyang Xu. “However, for its market development, operators need to reduce data-service charges, and vendors should provide more stable equipment and smart terminals. LTE deployment is likely to begin with hotspot mobile broadband access and then expand step by step.”

Currently, the most important element for operators is the technical maturity of terminals and LTE base stations. Challenges also arise from deploying such radio-access technologies such as IOT and SON, handling voice over LTE, and migration from existing networks.

Although standards for basic network functions are mature now, future migration and convergence in different radio-access technologies, such as GSM, UMTS, LTE, and CDMA, are flexible and are configured to work on alternative radio bands, which can create difficulties for vendors and future implementations. However, for basic functionality, commercial LTE networks can be deployed today.

Products and differentiators. ZTE describes its products as providing a future-proof end-to-end LTE/EPC solution for operators. The LTE system is based on a software-defined-radio (SDR) platform, which can be configured as a convergent multimode base station for different radio-access technologies (RATs). The EPC system, ZEPS, is based on an enhanced ATCA platform for the control plane and on advanced router platform for the media plane. ZEPS provides a multi-access mobile broadband access core network with high capacity and high performance, and can also help operators to overcome the mismatch of revenue and data-traffic growth with the ZTE Optimized Operation Management Solution (ZOOMS) by building an intelligent and value-added pipe with a user behavior analysis system (UBAS). Currently, ZEPS uses the company’s IMS 6.0 solution to support voice over LTE.

ZEPS includes the ZXUN xGW unified packet network gateway. Based on the 3GPP PCC architecture and UBAS, this gateway builds an intelligent and value-added pipe with a UBAS. The ZXUN xGW, which combines the SAE-GW and PCEF, has an embedded DPI function for service identification and policy enforcement. The PCRF is a decision-making entity for data-transmission policymaking and flow charging control according to information provided by the AF, SPR, and PCEF.

Additionally, the UBAS acts as a data-mining tool and assists in the analysis and development of high-value services. It analyzes the network in terms of network flow, protocol type, service type, user and so on, and presents the inquiry results in various formats, such as tables and various types of chart. Such results can support both network and user analyzes -- for example, the analysis of the most popular services used by different user groups.

— Tim Hills is a freelance telecommunications writer and journalist. He's a regular author of Light Reading reports.

Back to Page One: Introduction