In some ways, wireless networks are completely misnamed,
since the vast majority of wireless traffic rides traditional wireline copper and fiber facilities for most of its trip from originator to destination. But for the portion of the network that is wireless, data services growth on 3G networks has created a mobile backhaul problem in terms of transporting this traffic to or from the wireline network to the user.
For 4G Long Term Evolution (LTE) services, the new Evolved Packet Core (EPC) gives operators an architectural advantage for transporting wireless traffic. Yet for operators deploying LTE from a 3G service-base, evolving 3G mobile backhaul to EPC may be as big a planning issue as evolving 3G radio networks and handsets to 4G.
Transport infrastructure for 3G and 4G services can be divided into two categories:
- Mobile service elements with components that are aware of registration, mobility and service control aspects of mobile services;
- Transport/backhaul elements that provide connectivity between tower locations and service points.
Making sure mobile backhaul accommodates both 3G and 4G services
Both mobile service and transport/backhaul elements must be evolved in harmony so they can transition between 3G and 4G services
Mobile backhaul and 3G transport strategies vary depending on exactly what kind of 3G services are used (e.g., EDGE, HSPA, CDMA). In practice, however, most operators have deployed either time-division multiplexing (TDM) or asynchronous transfer mode (ATM) (AAL2) backhaul facilities to take advantage of their metro infrastructure and core technology.
This backhaul structure must be migrated to EPC during the evolution to 4G, and since fiber-feeding the tower sites is the preferred approach to 4G deployment, either parallel TDM/Ethernet or TDM over Ethernet Pseudowire Emulation Edge-to-Edge) is likely to be deployed. .
The question of integrating TDM backhaul for 3G with LTE/4G backhaul (which is also appropriate for High Speed Packet Access (HSPA) creates a question in the Evolved Packet Core's basic topology. EPC architects know that the logical architecture of an EPC connection in the data plane is tower (eNodeB)-to-gateway-to-service. The gateway in this context is the place where mobility registration and service control meet address assignment and data network connectivity.
Evolved Packet Core specifications, however, provide for the separation of the gateway into a serving gateway (SGW) and a packet data network (PDN) gateway, or PGW. This separation creates an EPC sub-network of tunnels where Quality of Service (QoS) and traffic management are more directly under the control of the service control logic (such as IP Multimedia Subsystem (IMS). This capability can be used to provide low-latency transport for TDM being transported over LTE facilities.
3G/4G Evolved Packet Core migration and integration planning
For migration planning, it is convenient to view the mobile service elements of 3G and 4G services by pairing elements roughly according to function. 3G circuit-mode connections used for voice can be simply tunneled or carried as noted above using parallel TDM/ATM or integrated pseudowires over Ethernet or IP. 3G packet traffic handling is where the tightest integration between the mobile service elements and backhaul strategies must be considered.
A good starting point is to consider the ultimate 4G logical model of mobile service elements and how they integrate with the 3G model. The Serving GPRS Support Nodes (SGSN) and Gateway GPRS Support Nodes (GGSN) relationship in 3G corresponds to the "trio" of Mobility Management Entity (MME)/SGW/PGW in LTE's EPC. The question, then, is whether the 3G and 4G elements are interconnected or integrated, which will depend on each operator's 4G architecture.
Mainstream network equipment vendors support three basic models for providing mobile service elements in Evolved Packet Core:
- Elements can be discrete nodes, with each logical EPC component representing a unique device.
- Elements can be fixed cards or interfaces on a router/switch device, so that logical EPC components are mapped not to their own devices but to specific packet metro/core elements.
- Elements can be "logical" and hosted by one of many switch/router or other service components in the network.
For any of these models, 3G and 4G functionality can be provided either independently or hosted in a single device. The latter solution is optimal where there will be considerable 4G deployment, where 3G elements are older (and thus represent less asset displacement cost), and where service evolution to 4G is expected to be rapid.
Where it's not feasible to replace SGSN/GGSN functionality with a dual 3G/4G/EPC node set (MME/SGW/PGW), the only option is to link the 3G elements with the 4G network to combine the traffic. Where integrated fiber backhaul connects tower sites with both 3G and 4G radio access networks (RANs), the use of integrated functionality for 3G/4G elements is much preferred because all traffic will emerge from backhaul at the same point.
Where operators are integrating 4G mobile elements into packet edge devices, it may be difficult or impossible to support both 3G and 4G missions on the same node because of constraints in 3G support by packet edge cards. Because of the flexibility that packet-edge hosting of EPC components offers, it is possible to host EPC components at the edge adjacent to the location of the 3G SGSN/GGSN that must be connected, and thus to reduce handling and latency.
Aiming for flexible Evolved Packet Core deployment
The most flexible approach would be to treat all of the Evolved Packet Core elements (MME, SGW, PGW) as logical entities that can be combined and hosted on available equipment in a variety of ways as network service demands evolve. This could allow operators to align the EPC components with current 3G elements and to create tunnels between 3G and EPC for packet traffic (UMTS Terrestrial Radio Access Network, or TRAN, traffic) where appropriate. As voice, data or all traffic evolves off of the 3G network, the location of the logical functions could be revised by changing the hosting points.
Many mobile operators and planners still think in terms of discrete devices when they think of mobile service elements, but that trend is reversing as operators understand the flexibility and operations benefits of having their packet edge devices host EPC roles. If that hosting is further enhanced by a "logical EPC" capability to permit rapid reconfiguration of the relationship between the EPC and the underlying metro/core network, the result is a structure that adapts not only to the evolution from 3G to 4G, but also to the changes in traffic and services needs that will inevitably come in a mature 4G market.
About the author: Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982. He is the publisher of Netwatcher, a journal addressing advanced telecommunications strategy issues. Check out his SearchTelecom.com networking blog Uncommon Wisdom.
This was first published in March 2010