Good Mobile Backhaul: There’s No App for That!

James P. Cavanagh

Mobile backhaul and wireless backhaul are two names for the same thing: the communications link between the practically ubiquitous cellular antennas on radio towers, buildings, highway signs, church steeples, fake Sequoyah trees and elsewhere that connect mobile subscribers to the carrier IP network and/or Internet. The backhaul should be the simplest part of the end-to-end link, and maybe someday it will be, but at the present time it is the biggest barrier to good, consistent end-to-end connections for wireless subscribers. And, at least so far, when it comes to good mobile backhaul system design there is no “app” for that. What largely characterizes mobile backhaul today is an odd balancing act between circuit and packet that could best be called a “truce” in the Circuit-Packet wars. But wasn’t that war already fought and won? Didn’t circuit get a full-on trouncing? Yes, at least for now, but sometime there is just no other option than to insert a SONET or SDH link between the two packet networks.

Why does, or should, anyone care? For the same reason that flyers who are loyal to one airline may find themselves flying on another airline: there is no other way to get to where you are going! When a wireless carrier enters or expands in a market they are often at the mercy of the dominant, or sometimes sole, local carrier and that local carrier’s bag of old circuit tricks with no clear incentive for ye olde carrier to abandon their ancient TDM ways and embrace the new packet craze.

Bridging the Gap: Mobile/Wireless Backhaul Technology Options

There is a virtual technological cornucopia of possible options but it is rare (and I’ve never seen a single, real-life case) that all of these technologies are available from a single provider in one, let alone multiple, markets. Let’s take a look at the “menu” and see what is possible.

[Traditional/channelized] T1: So, who uses T1 anymore? You might be surprised to find that there is still a LOT of T1 out there but you’d have to get your hands on proprietary “as built” drawings or circuit maps to prove it. T1 is still tariffed and available virtually everywhere and can provide 24 dedicated 64K channels. The good news is that the channels are completely separate from one another and do not require any complicated Quality of Service mechanisms in order to deal with mixed, multimedia traffic loads. The bad news is that each individual T1 channel is still limited to 64K which is the very minimum for modest QoS in today’s mobile multimedia applications.

[Non-channelized] T1: Take a traditional T1 which is divided into 24 channels of exactly 64K each, take down the dividers between the channels and what have you got? A non-channelized T1 with all of the narrowband lanes now merged into a single broadband channel of approximately 1 1/2 megabits per second. No single mobile device today can exceed that capacity and, in fact, as voice and video compression improve even more mobile subscribers’ data can harmoniously co-exist crossing the bridge from mobile wireless to terrestrial networks using this traditional, tried and true mechanism. Its small capacity, relative to the aggregate demand, makes it less than ideal, but like the frequent flyer who uses a non-preferred air carrier it is often the only way to get to the destination or it is the fastest way to get to the destination and fill the time gap while bigger, better solutions are being installed.

[Inverse-Multiplexed] T1: This is a solution which has been around for a long time but which I have rarely seen used with backhaul and no one can explain why. The idea here is that instead of taking numerous lower capacity links and joining them to go across a single, higher capacity link the inverse occurs: multiple lower capacity links are combined – and any subtle timing differences are resolved – so that a higher capacity link can be split across lower capacity links for transport and be recombined once they reach the other end of the lower speed links. It is very common to combine 4 T1s at 1.5 megabits per second each to get a single 6 Mbps connection, or 8 T1s for an aggregate 12 Mbps connection. This is one option that should be used more often when T1s are the only way available to get out to a radio tower, at least while waiting for higher speed connections to be available. This is also a plausible back-up plan for higher speed connections, especially “through the air” technologies such as microwave radio and free space optics.

[Traditional/channelized] SONET/SDH: This is the traditional phone company option, the younger, higher capacity sibling of T1 and T3. Just like T1 and T3 the bandwidth can be carved into dedicated, parallel paths as small as 64 thousand bits per second but into larger channels as well, often single channels as large as multiple T1s or even T3s. The good news is that bandwidth can be allocated and protected. The bad news is that in allocating and protecting bandwidth, and assuring the Quality of Service of the traffic traversing that bandwidth, some of the bandwidth is wasted. But bandwidth is less expensive than ever and the bandwidth cost <-> quality tradeoff is worth the money in certain applications.

[Non-channelized] SONET/SDH: In this more modern version of SONET and SDH there are no barriers: all traffic travels in a single big channel with frame or packet structures being used to multiplex and demultiplex the units of information. For basic Constant Bit Rate (CBR) traffic, the overhead of Layer 3 IP packetizing or Layer 2 framing (such as Ethernet) is very wasteful of bandwidth and introduces additional delay but there is very little CBR traffic in the mobile backhaul of today. Virtually all applications put their information content natively into IP packets. IP packets work very well with non-channelized SONET/SDH, as do the Layer 3 Quality of Service mechanisms built into the Type of Service (TOS) bits of IP. This is true whether the non-channelized SONET or SDH bandwidth is treated as managed bandwidth, often because of the cost and other limitations of SONET or SDH bandwidth over microwave or the non-channelized SONET or SDH bandwidth is treated as “big bandwidth”.

Carrier Ethernet: Much has been said about Carrier Ethernet and it is looked on by many to be the ultimate backhaul solution. But things are not quite that simple. At the most basic level perfectly good IP packets are put into Carrier Ethernet frames for transport, usually at the entry point into the mobile backhaul connection and are taken off somewhere after the frames exit the mobile backhaul. In this case the QoS and prioritization mechanisms are embedded into the VLAN IDs and the VLAN IDs are used to multiplex and de-multiplex Ethernet frames onto and off of the shared channel. The VLAN IDs are often only meaningful on the backhaul, which is problematic in identifying the connections and delivering the promised QoS.

SONET/SDH Transport: If SONET or SDH is chosen as the framing mechanism then it is necessary to use a transport which enforces rules that are required for successful transport, such as timing and framing. This may mean either SONET/SDH radios which are usually microwave or SONET/SDH terrestrial fiber. Most traditional “Ma Bells” usually have a lot of this available and, very often across ring or mesh architectures that come with re-routing and reliability built in.

PACKET/FRAME Transport: Another bit of good news is that non-channelized SONET/SDH running over SONET/SDH transport is a perfectly acceptable way of getting IP packets of Carrier Ethernet frames across the back-haul and is often a solution that starts as an “interim solution” and is there, working reliably, for years and years. A second option is Ethernet radios which use native-mode Ethernet transport. While SONET/SDH transport may be used for SONET/SDH frames, Ethernet frames and/or IP packets Ethernet radio can only be used with Ethernet-framed traffic, which can be native Ethernet or IP but does not include SONET or SDH. It is also true that the Ethernet radio market is still very young and dynamic with all the attendant flexibility and flux.

Other Wireless Options: There are also other wireless options to consider for transport, each of which can be outfitted with either SONET/SDH or Ethernet interfaces or, in the case of a transition, with both, such as Free Space Optical (FSO). And, it is often a good idea, if the connections being carried across the mobile backhaul are critical, that a wireless backup be provided for the terrestrial connections and vice versa. And, in a closer-to-perfect world, half of the live traffic would run over the terrestrial and half of the live traffic would run over the wireless all of the time.

Conclusion

What should you do now? If you are a regular, ordinary wireless subscriber you are pretty much stuck with what your carrier has chosen, but you do have a choice of carriers even though issues such as this will very likely not influence you significantly enough to make you change carriers. If, on the other hand, you are shopping for a new carrier or if the wireless communications link is of life or death importance then it is worth diving in deeper and understanding the resilience and availability of your lifeline service.

If you are a wireless carrier trying to create the best possible and most consistent Quality of Experience for your users then you should build a proper Service Level Agreement and monitor it as if your life depends on it, and in a way it does. Have an internal SLA with stricter requirements than those of the customer’s SLA and make sure you meet or exceed your own internal requirements. Also, report your compliance results to your customer regularly and earn all the kudos and praise to which you are entitled. If you don’t tell them the good news you’ll only hear from them with bad news. And, maybe someday, just like circuit network design and packet network design there will be an app for backhaul design. Or, maybe it will be improved until it is just another link in the end-to-end puzzle.

Editor’s Note: Besides 30+ years of experience that includes hands-on engineering, consulting, and training on a wide range of networking technologies, Jim Cavanagh, a Principal Member of our Telecommunications Faculty, is our resident expert on mobile backhaul as well. He has taught numerous classes for clients using a wide range of backhaul options — including traditional SONET/SDH, multiple broadband technologies, satellite, and other wireless technologies — to get the multimedia traffic from their users’/clients’ hands to the ‘net. He has instructed backhaul engineers in the proper set-up of multiple backhaul devices; advised large multinationals, government, and public safety agencies on backhaul security and the use of backhaul as a part of VPN and Cloud-based networks; documented backhaul best-practices; helped establish SLAs and service descriptions; and even written backhaul marketing materials.