After all, we are just trying to transport a simple Ethernet frame from the base station (BTS, NodeB, eNB) to the base station controller (BTS, RNC, S-GW, ASN-GW).  As the cellular industry moves toward "all IP networks", they are laying the groundwork for an effective unified backhaul solution.  The solution that seems to have universal acceptance is the use of "Metro Ethernet".  That is the deployment of a carrier grade Ethernet network solution capable of supporting both the backhaul needs of today (e.g. T1/E1 TDMs, ATM, HDLC, IP) as well as backhaul needs to be presented by 4G (and 5G) networks of the future.

To appreciate how complex this "simple" task of backhauling Ethernet is, you need to be a fan of the Rube Goldberg Machine.  A Rube Goldberg machine is a deliberately over engineered apparatus that performs a very simple task in a very complex fashion, usually including a chain reaction.  The picture below is an example of such a machine which has the simple objective of "de-mothing" the clothes in the closet.   

To see the parallel with the Rube Goldberg Machine, consider the chart below depicting the various solutions to address Ethernet backhaul.   This is a characterization of what Metro Ethernet Forum calls the Ethernet Services Model.  This is a three layer model with the applications supported (e.g. TDM, ATM, etc.) at the top.  The middle layer represents the different standardized services supported by Metro Ethernet.  To date that have three services types; 1) a standardized point to point services type called E-LINE, 2) a standardized multipoint to multipoint services called E-LAN, and 3) a rooted point to multipoint services type called E-TREE.

The third layer is the killer.  It is called Carrier Ethernet Transport (CET) and describes various transport solutions that couple with Metro Ethernet Applications and Metro Ethernet Services, to provide a Metro Ethernet backhaul solution.  By the way, the terms "Metro Ethernet" and "Carrier Ethernet" are sometimes used interchangeably.  In fact the "Metro Ethernet" Forum in the organization that defines the "Carrier Ethernet" standards.

The CET solutions can be viewed is three sets.  The solutions on the left are the "Multi-technology" solutions.  These are the Rube Goldberg Solutions.  The solutions on the right include a Pure Ethernet Solution (Provider Backbone Bridging - with Traffic Engineering - PDD-TE) and an MPLS solution (Multi Protocol Label Switching - Transport Profile - MPLS-TP).  These are the mode straight forward solutions. 

To appreciate the Multi-technology alternatives, consider the possible use of Resilient Packet Rings - RPR (thought by some to become the SONET/SDH of the future).  There are numerous ways RPR can be deployed but one realistic contender is this:

• Embed the Ethernet frame being transported inside an MPLS header to get an MPLS frame
• Embed the MPLS frame inside an RPR header to get an RPR frame
• Embed the RPR frame inside a Generic Framing Protocol (GFP) header to get a GFP frame. Note: GFP used to map packetized data into synchronous frames
• Map the GFP frame into a SONET Payload Envelope (SPE) which is then encapsulated with a SONET header to get the SONET frame. Now every 125 milliseconds, SONET frames will be carrying your Ethernet data

A natural question is, "If the PBB-TE solution or the MPLS-TE solution is a better solution, why not just go there initially?".  There are actually several good answers to that.  First, neither PBB-TE nor MPLS-TE are yet approved standards.  The PBB-TE proposed standard is the closest is finalized.   A search of the web suggests that MPLS-TE may be the preferred between the two alternatives.

A second reason that carrier do not instantly jump to the "better solution" is that new solutions require new commitments to network equipment and software.  Migration to new networking solutions in large networks takes years of planning and preparation.  As cumbersome as some of the Multi-technology solutions may seem, they provide the carrier an opportunity to begin migrating to Ethernet backhaul today using today's equipment.

A third reason not to jump to a new backhaul solution involves the natural evolution expected toward an all-IP network.  This will happen over time.  The 4G solutions like Mobile WiMax and Long Term Evolution (LTE) claim to be capable to supporting up to 300 Mbps of traffic from the cell site.  Few doubt that these claims will be eventually realized, but it will take some time. 

Finally a primary tool to be used by the carriers to extend the life of their current network infrastructure, and grow the backhaul in proportion with  increase in demand is "Bundling".  Bundling is techniques where multiple "links" are "bundled" together os appear as a single link having larger capacity.  Consider the fact that E1 has the capacity to transport 2 Mbps.  If this is insufficient, the standards define the next higher transport container to be an E3 with a capacity of 16 E1s (i.e. 32 Mbps).  Bundling with a technique called MultiLink Point to Point Protocol (MLPPP), would enable the carrier to bundle 5 E1s to create a "personalized" container of 10 Mbps.  On other words, the carrier could grow with the traffic rather than needing to make a quantum jump in the deployed capacity. 

As one person once said, there is a light and the end of the tunnel for Ethernet Backhaul (e.g. PBB-TE and MPLS-TP) and it is not a train!!