 |
High Bandwidth Metro Wireless Backhaul - Ethernet vs SONET Capital Cost Performance Comparison
Erik Boch, CTO & VP Engineering , Dragonwave
Introduction
Wireless access services are continuing to find escalating market opportunities. The addition of new services places an ever-increasing demand for increased access bandwidths in wireless networks, both fixed and mobile. The industry has responded with a variety of technological solutions, including bandwidth-enhancements to existing networks (i.e. HS/US-DPA and LTE for GSM, EVDO for CDMA) as well as new technology build-outs (i.e. WiBro, WiFi, WiMax).
Along with the increased bandwidth in the access layer of these networks, comes a need for high bandwidth, high performance backhaul in order to connect the [distributed] access layer infrastructure aggregation nodes (i.e. mobile base-stations) back to the metro core. The large bandwidth demands drive a need for new network solutions ... beyond traditional TDM leased-lines. As most of this bandwidth is data-centric and Ethernet-based, Ethernet-based backhaul networks are naturally appealing due their inherent efficiency and cost effectiveness. Further, wireless backhaul networks are highly appealing due to their cost effectiveness and rapid deployability (compared to other high bandwidth options, namely fiber optics).
Although native wireless Ethernet backhaul of Ethernet-based traffic would nominally provide obvious efficiency gains, it is perhaps not so obvious that they are also more cost effective than other options (SONET-based solutions for example)
This paper examines a cost-performance comparison between Ethernet wireless metro backhaul and its SONET counterpart. The primary focus of the analysis is the capital costs associated with these metro backhaul networking options. The analysis compares and contrasts these using information gathered from actual metro backhaul network designs.
The Importance of High Performance Backhaul?
In the past, mobile services have steadily increased their bandwidth demands on mobile networks and this is expected to continue into the future. Figure 1 shows the general trend that the industry is seeing along with the access technology solutions that the operators have employed to meet these demands. Looking at the current and future access technologies, WiMax and LTE are expected to play vital roles in the delivery of mobile services. Both of these technologies rely primarily on packet-based (Ethernet) backhaul, in contrast to the majority of current backhaul which is low bandwidth TDM in nature.
Figure 1 – Mobile Services Data Rate Demand Over Time[1]
The basestation technologies that were deployed in the past were designed primarily for the delivery of mobile voice services. TDM transport (N x T1 oe E1 circuits) was typically used for backhaul. It’s general availability and cost effectiveness made it a logical fit for this segment of the network. However, as the services (and associated traffic flows) become more data-centric, the backhaul networks began migrating toward technologies that can more cost-effectively transport this kind of traffic. As illustrated in Figure 1, the trend is toward Packet-based Ethernet transport in the metro backhaul layer.
Similarly to Figure 1, Figure 2 illustrates trends in the deployment of backhaul technologies. With the higher data rates and higher data content, native Ethernet (packet based, connectionless) backhaul is on the rise and its SONET (synchronous) counterpart is declining. Also worthy of note is the fact that wireless/microwave technologies are already the predominant backhaul deployment vehicle, representing ~ 50% of deployed mobile base station sites today.
Figure 2 – Changes in Transport Technologies Being Deployed in Mobile Wireless Networks[2]
Technically, using Ethernet to carry data traffic makes sense since the transport efficiency is naturally high when the traffic payload has a high percentage of connectionless applications/sessions in it. Additionally, data applications tend to drive a heightened demand on the networks’ ability to scale (up-speed) since new data applications can be [are] generally more bandwidth-demanding. However, transport efficiency and scalability are only the technical portion of the solution that network operators are after. The other important element is cost-effectiveness.
As the traffic has shifted toward data-centricity, mobile operators are increasingly confronted with reduced revenue (see Figure 3). To combat the trend shown on the left portion of this figure, operators are increasingly looking to the next generation network solutions to achieve the performance shown on the right.
The improvements in the network costs are associated with many aspects of network design, equipment, installation, operation, maintenance and [many] other cost-of-ownership elements. A critical part of the whole network performance and cost puzzle is the metro backhaul layer. Wireless backhaul technology is increasingly being seen as an ideal choice for minimized total cost of ownership in forward-looking network builds. This technology allows for cost-effectiveness on many fronts;
Figure 3 – Traffic Growth vs Revenues[3]
Wireless Backhaul Networks – SONET vs Ethernet
Mobile networks are at a cross-roads …. existing synchronous networks which are voice-centric must grow a huge capability to handle the onslaught of data…..and newer packet-based [data-centric] networks must often host the backhaul burden of legacy TDM transmission that backhauls current [revenue-generating] mobile voice traffic. The latter problem can be dealt with through the use of Pseudo-Wire (PW) technologies which employ advanced timing-over-packet protocols to distribute high resolution timing information throughout the packet network allowing them to emulate synchronous network behavior[4] .
At a high level, the metro backhaul views of these network topologies are similar is shown in Figure 4. At this level the differences are rather subtle and one could see them as being equivalent. Digging into slightly more detail, the backhaul “cloud” shown in this figure may take the form of several different wireless backhaul network structures. Figure 5 illustrates some of the possible sub-circuit topologies. These would tend to get step-and-repeated in order to achieve a metro-wide backhaul network solution.
The conventional Daisy-Chain (a) and Ring (b) can be realized using either SONET or Ethernet transport but the constrained mesh (c) is only addressed well by Ethernet. Ring and Mesh topologies offer the operator several attractive benefits;
Additionally, the constrained mesh topology offers x2 bandwidth scaling over the ring. This is achieved by the added “cross-bar” link in the constrained mesh topology.
Figure 4 – SONET and Ethernet-based Metro Backhaul Network Architectures
Figure 5 – Typical Metro Wireless Backhaul Network Sub-Circuit Structures
Summary – Wireless Backhaul Network Costs
In addition to the efficiency-related cost benefit of carrying the [larger] data-centric payload on native Ethernet backhaul, Ethernet-based networks also enjoy several other CAPX benefits. Carrier-class Ethernet point-to-point radios and networking switching equipment are generally much lower cost than SONET equipment due in part to the core IC technology being employed[5]. Ethernet components have the benefit of being attached to the global high volume enterprise/LAN market that SONET/TDM do not really participate in. This not only changes component volume-related pricing, it also changes the operating and margining mentality of the entire product supply chain. This latter element is important because it drives a culture of cost sensitivity that allows the supply of cost-effective, yet high performance products.
In response to the desire to consume Ethernet cost-effectiveness in carrier networks, Ethernet has evolved from a feature/functionality perspective to become suitable for carrier deployments. Carrier-oriented standards organizations[6] have been, and continue to be, instrumental in evolving Ethernet technology to the point where Ethernet can be considered “carrier grade”.
Bandwidth scalability is also a key performance attribute that can also drive significant CAPX increments in the network, Ethernet enjoys highly monotonic scaling costs allowing operators to “pay-as-you-grow” their network scaling costs. Ethernet, being data centric in its origins, is well acquainted with the varying demands that data-users can place on networks. As a result a great deal of flexibility is present in most of the product designs.
Figure 6 – Comparative Costs of Ethernet and SONET Wireless Metro Backhaul Networks
The Ethernet implementation is typically 30% – 50% lower in cost depending upon whether Psuedo-Wire is needed for TDM/legacy backhaul and also depending the type of network switching functionality selected. L2/L3 and PbT –based switching tend to be the lowest cost whilst MPLS tends to push costs upward somewhat, offering functionality which is closest to conventional switched circuit networking.
Regardless of Ethernet switching technology selected, high performance, high bandwidth carrier-grade Ethernet radio equipment[7] forms an integral part of the overall backhaul network solution. These Ethernet radio systems can include a full suite of management, QoS and OA&M features making them directly suitable for inclusion into a carrier-grade metro Ethernet backhaul solution. [1] Alcatel-Lucent, IWPC, Milan, Jan. 2008 [2] Ericcson, IWPC, Milan, Jan. 2008 [3] Alcatel-Lucent, IWPC, Milan, Jan 2008. [4] see suppliers such as www.dragonwaveinc.com
[5] Ethernet ICs typically enjoy significantly lower costs due to the high volume LAN market which they participate heavily in [6] i.e. IEEE, Metro Ethernet Forum (MEF) [7] see suppliers such as www.dragonwaveinc.com
Erik Boch is the CTO and VP of Engineering for DragonWave Inc. (http://www.dragonwaveinc.com), he holds a Masters degree in Electrical Engineering from Carleton University in Ottawa and is a registered professional engineer. Erik can be reached at eboch@dragonwaveinc.com.
|
|
