By Ralph Santitoro
By combining the best attributes of connectionless Ethernet and SONET/SDH, connection-oriented Ethernet is ideally suited to mobile backhaul networks, providing highly efficient bandwidth utilization and scalability with stringent QoS performance and high availability.
Demand for higher-speed wireless data services continues to surge. Just as Wi-Fi has become widespread in homes and businesses, the mobility brought by high-bandwidth 4G wireless technology is expected to become perhaps even more ubiquitous than Wi-Fi.
New 4G wireless technologies, specifically Long Term Evolution (LTE), will place unprecedented demands on the wireline telecommunications network infrastructure. It is this infrastructure that backhauls traffic from cell sites to mobile switching offices, which connect to the Internet and voice networks. The current 3G backhaul network infrastructure, which typically consists of a few to several T1 or E1 circuits or microwave radio channels to each cell tower, may be adequate where each subscriber might use hundreds of kilobits per second of bandwidth. However, with 4G services, a mobile subscriber will be able to use several megabits per second of bandwidth (more than a T1/E1 of bandwidth per subscriber). The current T1/E1 infrastructure is clearly inadequate for such services, since multiple 4G subscribers may need connectivity to a given cell tower.
In addition, the vast majority of mobile backhaul networks based on SONET/SDH technology are optimized for legacy voice services, but less so for 4G packet-based data services, due to the channelized structure in 50-Mbps (SONET) or 155-Mbps (SDH) increments. With flat-rate data service plan pricing, mobile operators require their backhaul network infrastructure to scale to meet the new bandwidth demands of 4G services, while also providing maximum bandwidth utilization of that infrastructure. Mobile operators do not own the backhaul network in all the markets they serve. To maintain or increase profitability, they must carefully manage their monthly recurring operations costs for such leased backhaul services.
Mobile operators are focusing on their backhaul infrastructure because this is an area that can impact opex in the area of operations, administration, and management (OAM). The traditional TDM-based mobile backhaul network inherently provides deterministic connection performance in addition to the lowest possible latency, highest network availability and security, with essentially no data loss. The wireless standards were designed with these types of networks in mind. Therefore, a packet-based network architecture better optimized for 4G packet-based services must also meet the stringent requirements of these legacy TDM networks.
Ethernet to the rescue—but which implementation?
An Ethernet-based network provides the bandwidth scalability and highly efficient aggregation required for all-packet-based 4G services. Ethernet is augmented with a set of companion technologies to improve reliability, minimize packet loss, and provide the low latency needed to meet the stringent requirements of mobile backhaul networks. However, there are several transport technologies over which Ethernet can be delivered: native Ethernet, SONET/SDH, or MPLS. For each of these, Ethernet can be further differentiated between connectionless Ethernet (CLE) and connection-oriented Ethernet (COE).
CLE in service delivery networks is the fundamental Ethernet switching (bridging) technology defined by IEEE 802.1ad Provider Bridging (commonly referred to as Q-in-Q) and further advanced in IEEE 802.1ah Backbone Provider Bridging (often referred to as MAC-in-MAC). CLE provides the fundamental benefits of Ethernet such as packet aggregation and statistical multiplexing. However, it cannot meet stringent quality-of-service (QoS) performance and reliability (OAM) requirements without augmentation via additional technologies. This is one of the reasons COE technologies have been introduced.
COE technologies provide the deterministic and precision QoS (packet loss, packet latency) and reliability inherent in SONET/SDH, combined with the packet aggregation and statistical multiplexing benefits of CLE. By combining the best attributes of CLE and SONET/SDH as illustrated in Fig. 1, COE is ideally suited to meet mobile backhaul requirements, namely highly efficient bandwidth utilization and scalability with stringent QoS performance and high availability.
COE critical to mobile backhaul
The mobile operator or backhaul network provider needs the ability to easily and efficiently scale bandwidth to support a growing subscriber base for high-speed 4G data services. A highly reliable backhaul network is necessary because of the large quantity of mobile subscribers that can be affected in an outage and because many of the protocols were designed to operate over a TDM infrastructure that features sub-50-ms protection/restoration performance.
COE meets these requirements because it is a high-performance implementation of Carrier Ethernet. As shown in Fig. 2, COE supports the Metro Ethernet Forum’s five attributes for Carrier Ethernet. COE provides deterministic QoS through the provisioning of explicit, traffic-engineered network paths with consistent delay, delay variation, and loss profiles. This ensures the best quality of experience (QoE) for mobile users and provides the requisite QoS for OAM traffic such as clock synchronization, telemetry, and mobile backhaul network performance measurements.
Mobile backhaul networks also must ensure that critical services and applications receive the necessary bandwidth and enable the mobile operator to maximize bandwidth use of lower-margin, flat-rate priced data services. Through a connection admission control (CAC) process, COE reserves bandwidth per Ethernet virtual connection (EVC), providing precise bandwidth guarantees via a committed information rate (CIR) while enabling use of excess bandwidth and oversubscription through an excess information rate (EIR).
COE implementation approaches
COE can be implemented using an Ethernet-centric or MPLS-centric approach. Each approach may meet the technical requirements for mobile backhaul networks, but there are significant differences from an operations perspective that result in significant opex differences. Since opex is a critical focus area for mobile backhaul network providers and mobile operators, these operations differences are often the deciding factor in technology selection.
Technology selection also depends upon the expertise and experience of the network engineering and operations staff. Another important factor is whether the mobile backhaul network is only supporting 3G or 4G packet-based services, or is also supporting legacy 2G or 3G services using MLPPP (EVDO) or ATM (HSPA) encapsulation technologies over T1s or E1s.
In the latter case when legacy 2G and 3G services are also present, there are several potential approaches depending upon the business drivers. Assuming an existing SONET/SDH-based backhaul network infrastructure, opex is contained for the existing one or two T1/E1s used by 2G services because 2G service bandwidth growth is flat or declining. With 3G data services, the rapidly increasing bandwidth demands can make backhaul over T1s/E1s impractical given the incremental monthly recurring costs for each leased T1/E1 and the quantity of T1s/E1s needed to provide the necessary bandwidth.
Manufacturers of 3G base stations often have upgrade options for these platforms that provide an Ethernet interface in lieu of the T1/E1 interfaces. This is the simplest technical approach to evolving the mobile backhaul network to Ethernet. Without an Ethernet interface, a generic interworking function (GIWF) is required to circuit-emulate or encapsulate the ATM or MLPPP over bonded T1s/E1s into Ethernet. This adds a considerable amount of complexity when compared to migrating the base station interfaces from T1s/E1s to Ethernet.
Since SONET/SDH natively and most efficiently accommodates T1s/E1s for legacy 2G services that are not growing, optimizing the existing SONET/SDH backhaul network using COE for packet-centric 3G and 4G services makes a lot of sense. As 2G subscribers eventually migrate to 3G or 4G services, the need for TDM is eliminated and the SONET/SDH transport layer in the backhaul network can then migrate to COE over fiber as shown in Fig. 3.
For backhauling strictly 4G services, the technology selection is simpler because 4G is solely IP-based and 4G base stations only support Ethernet interfaces. In this case, the mobile backhaul network can use COE over fiber. For existing SONET/SDH-based backhaul networks, COE can be used to provide a highly efficient, packet-optimized solution over SONET/SDH for 4G services.
When implementing an Ethernet-centric COE approach for mobile backhaul networks, OAM is simplified because only Ethernet-based OAM tools are required. For example, when implementing Ethernet-centric COE, all link and service faults (loopback and linktrace) and service performance (frame delay, inter-frame delay variation, and frame loss) are managed using the same set of OAM tools because only one Ethernet data plane layer needs to be managed.
Connection-oriented Ethernet technologies were designed to provide deterministic and precise QoS, security, and reliability of SONET/SDH with the Layer 2 aggregation, scalable bandwidth, and statistical multiplexing benefits of Ethernet. These attributes make COE ideal for 3G and 4G mobile backhaul network applications. COE can be delivered directly over fiber or it can optimize an existing SONET/SDH infrastructure for 3G and 4G Ethernet-based services, while coexisting with legacy TDM-based 2G services. Finally, COE enables mobile operators to minimize opex for high-growth 3G and emerging 4G services while by providing the lowest cost and simplest OAM when compared to other COE implementations.
Links to more information
LIGHTWAVE:Connection-Oriented Approaches Complete the Ethernet Revolution
LIGHTWAVE:Operators Tackle Mobile Backhaul with Optical Ethernet
LIGHTWAVE:Changing Paradigms Affect Cellular Networks
Ralph Santitoro is a market development director at Fujitsu Network Communications (www.fujitsu.com/us/services/telecom/).