"Wireless networks" are named for the so-called "last mile," the portion that delivers the connection to a handset. For the rest of the information path, the normal practice
The fact that an evolved packet core replaces backhaul may give the impression that LTE tower connections form a completely separate network. In some markets, that may well be; where wireless services are built because there is limited or no existing wireline infrastructure (the case in evolving economies, for example). In major industrial economies with mature wireline networks and evolving wireline consumer broadband deployment, however, it's far more likely that EPC will be intimately linked with metro/middle-mile deployments.
Almost all Tier 1 operators believe that to be true, and require such integration in their LTE planning. It is fortunate that, by design, EPC facilitates this integration with wireline infrastructure, though just how it comes about may be hard to wrestle from the standards.
Three-layer LTE architecture highlights evolved packet core's purpose
For purposes of infrastructure planning, it's probably best to visualize an LTE deployment as being a radio network, a control layer and an evolved packet core layer.
- The radio network consists of the LTE towers that create the subscriber coverage.
- The control layer manages the registration of handsets, the identification of customers and the management of the handset relationship to the data infrastructure.
- The evolved packet core layer is the link between a mobile user and the fixed metro/middle-mile transport facilities, and provides session, mobility and quality-of-service (QoS) management, enabling operators to connect users to applications in their service delivery environment, on the Internet and on corporate networks.
When a wireless user turns on an LTE handset, it is authenticated by a service control function such as the IP Multimedia Subsystem (IMS) through a facility called the Mobility Management Entity (MME).
In theory, an evolved packet core can be created through any combination of IP and Ethernet, and there's a good chance that most EPC deployment will be an adjunct to the metro or middle mile networks being built. In fact, much of the functionality of the serving gateways and packet gateway could be embedded in an edge interface, and most vendors will likely offer that solution. Edge-hosted serving gateway functionality is especially logical if the LTE towers are served by discrete fiber, something many operators are considering because of the need for high bandwidth. Fiber trunks could then terminate on their own dedicated serving gateway.
Evolved packet core planning focuses on traffic handoffs
In planning an evolved packet core, it's smart to consider the pattern of handoffs that can be expected. Shifting a user to another tower location will change the traffic pattern somewhat, but the change will be less dramatic if cell towers that form a logical community serving users who migrate among them are connected to serving gateways located in the same device or rack. That will reduce traffic variability arising out of normal user movement between cells.
For evolved packet core applications, it is absolutely critical that the signaling exchanges (MME, SGW) and the service control components (like the HSS in IMS) are reliable. Signaling path failures will compromise services, and latency in managing handoffs will disrupt the bearer channel and interfere with user conversations or experiences. These signaling channels need not be carried on the same routes as the bearer channels, and while diverse routing may not be valuable because both bearer and signaling connections are needed for stable service, it may help to reflect priority handling of signaling paths.
It's important to remember that evolved packet core bearer channels are essentially tunnels that create an independent set of data paths that must be traversed before traffic is subject to normal IP routing. The location of the packet gateway relative to the serving gateways is the last consideration in evolved packet core design. Where the two are "close" in a network topology sense, traffic jumps off the evolved packet core and onto the IP service network close to the cell sites. That can be very efficient if there are many local servers and service elements accessed through the LTE network. It can mean having more packet gateway points, however, and less efficient 4G traffic aggregation. Centralizing packet gateway locations will mean managing bearer channel paths more explicitly because they're longer, and all traffic generated by the users of the network will have to be carried some distance before it can be connected to any server or experience source.
4G puts apps 'on the road,' stressing metro aggregation resources
A final point in evolved packet core design is the relationship between the EPC and other metro wireline applications, including any that arise from femtocell and fixed mobile convergence (FMC) deployment. A 4G network of any sort encourages users to take traditionally fixed applications and experiences "on the road," and thus it may be important to determine both LTE and wireline traffic patterns associated with resources -- especially content resources -- when designing the evolved packet core and the connectivity of LTE towers. Because there are likely to be many more LTE towers than central offices in a given metro area, LTE will eventually become the primary user of metro aggregation resources, and this shift will need to be reflected in the broader area of metro and middle-mile aggregation network design.
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.