Jarich: LTE TDD - Looking back to the birth of a standard

Current Analysis Peter Jarich

     Peter Jarich

We all knew the 5G pendulum was going to swing back pretty hard post-Mobile World Congress this year. Sure enough, come mid-year 2014, it's suddenly fashionable for people in the telecom industry (media, analysts, etc.) to push back on the 5G hype. You probably know the messages I'm thinking of: "There are still plenty of 2G networks supporting M2M and voice in operation." "3G will be with us for years, so keep investing in those networks." "LTE and LTE-A are more than capable enough to support millions upon millions of people watching million upon millions of streaming video sessions."

Yet, with the NGMN getting ready to release a white paper on 5G, it seems that the "too much, too soon" message might not be fully understood by the people who matter most--those likely to actually deploy 5G networks someday. Are you surprised? You shouldn't be. After all, it was 2006 when the NGMN released its white paper on "Next Generation Mobile Networks Beyond HSPA and EVDO" specifying what key mobile operators thought LTE should look like… and that was two years after the 3GPP began work on LTE at a RAN Evolution Workshop in Toronto, Canada.  

That's right, the 3GPP's work on LTE is a full 10 years old!

As we get ready to celebrate LTE turning 10, it's worth asking, "What type of birthday present do you get a mobile broadband standard?" How about a trip down memory lane? It might not be a great gift, but it should allow us to think about the role TDD plays in LTE, how TDD fit into the way the LTE standards were put together way back when. 

Looking back at the initial thinking about LTE as an evolution from 3G technologies, two types of requirements were outlined. Some were explicit. Others were implicit. Heck, the NGMN messaging was little more than a "wish list" for what LTE standards should include. Nonetheless, taken together, they tell a straightforward story.

The explicit requirements for LTE include a mix of features we all take for granted, along with some others that, while obvious, might not be top of mind.

-        Radio and Data Performance. It's only natural to think of new technology evolutions in terms of "speeds." Here, the operators in the NGMN knew just what they wanted: peak uplink data rates of 50 Mbps in a 20 MHz channel; peak downlink data rates upwards of 100 Mbps in a 20 MHz channel; 10ms latency in the RAN and core; spectral efficiency of at least 3-5X that of HSPA or EV-DO (with 6-8X being preferred); seamless mobility with broadcast and multicast support; support for at least 60 VoIP sessions per cell/MHz. It was a long list, and not aligned exactly with 3GPP requirements; the 3GPP, for example, called out spectral efficiency of 3-4X R6 HSPA on the downlink, and 2-3X on the uplink. Regardless, it set out what end users and network planners alike could expect.

-        Fragmentation Limitation. With LTE expected to be the migration path for mobile broadband operators of all stripes (expected by non-WiMAX supporters, at least), it was critical for the technology to be "inclusive." This meant interworking with legacy (GSM/HSPA) and even non-3GPP (WiMAX, WiFi, CDMA2000) technologies in the RAN, core and terminals. It meant supporting diverse spectrum allocations, channel bandwidth and duplexing schemes. It meant very basic things like support for network sharing, roaming and unified network management.

-        Complexity Limitation. Technical details are nice: they help the people building a standard to know exactly what's expected of them. Sometimes, however, simpler is better. That's why I love the 3GPP's 2008 recap of LTE requirements on the topic of "Complexity." It is two bullets: minimize the number of options; no redundant mandatory features. It's only natural that recommendations on complexity be simple, right?

Returning to these requirements (and the dozens of others that were outlined in the run-up to LTE getting commercialized), several others are clearly implied. LTE was to be an evolution from (and integrated with) 3G. Whether in terms of IP networking, network simplicity or spectral efficiency, cost reduction was critical. Playing into the cost reduction notion were two others: the scalability of the LTE ecosystem and the ability of operators to use any spectrum assets they could get their hands on.

Great. But what does this all say about TDD and FDD?

Both the explicit and implicit requirements for LTE as a standard point to the need for TDD and FDD support as part of a single standard. TDD and FDD were explicitly called out in terms of leveraging diverse spectrum assets. TDD was called out in terms of duplexing flexibility. The notion of LTE as a 3G migration path implied TDD support; recall the 3G technologies like UMTS TDD and TD-SCDMA that operated in TDD spectrum. And, of course, any hope of maximizing the scale of the LTE ecosystem required the technology to include both duplexing schemes. If nothing else, network and device vendors focused on regional 3G technologies (like TD-SCDMA) could now look forward to bigger, global opportunities. Likewise, vendors which chose to ignore those regional technologies would have an easier time selling into markets like China.

In other words, TDD was an integral part of LTE thinking from the outset--not a separate standard, not a "corner case," not an exception. More than just co-existence, however, it was clear that harmonization and/or convergence was the goal. But does this mean that they are one in the same? A single, unified standard with a single, unified ecosystem? Yes and no.

In explaining how TDD works, the analogy of a roadway is probably overused. FDD is represented by a two-lane highway. TDD, then, can be thought of like a one-lane bridge supporting traffic in both directions thanks to a stoplight regulating traffic in time. Different operations, sure. But, ultimately, each is a road serving a similar purpose. 

And yet, each comes with different user expectations and different infrastructure demands--much in the same way that LTE TDD and LTE FDD base stations are distinct entities and the number of user devices supporting LTE TDD is still dwarfed by the universe of FDD LTE devices.  LTE as a foundation provides clear scale benefits accruing to both, but we're also seemingly dealing with a case of "separate but equal" treatment in terms of the ecosystem. The question is whether this a necessity or something that we can expect to change with them. That's a great question to consider for next month!  

Peter Jarich is the VP of Consumer and Infrastructure at Current Analysis. Follow him on Twitter: @pnjarich.

Read more on