A 3GPP Standards Development Primer: Corrections in Perpetuity
==========
The following is an excerpt from Deutsche Bank's May 2002 report: "The Rise of the 3G Empire"
3GPP—The Long and Winding Road Keeps Getting Longer and a Bit More Treacherous
Members of the 3GPP include organizations such as ETSI, individual members and market representation partners such as the GSM Association. Included with the organizational and market representation partners are virtually every major OEM, a litany of smaller companies, regulatory agencies and carriers. The 3GPP, like the 3GPP2, is divided into technical specification groups, or working groups, that write the technical standards for their respective areas (e.g., GERAN, Core, RAN, Terminal). Once these standards are written, the 3GPP endorses the standard and submits it to the ITU.
One aspect of 3G that is often missed by the popular press is the concept of releases, a concept that also applies to 2G and 2.5G networks. 3G, in this case UMTS, is not one release, but a series of releases that builds upon the previous releases by offering fixes and adding enhancements to earlier releases. Initially, releases were noted by year (e.g., Release ‘99, Release ‘00); however releases are no longer tied to the year in which they are finalized. Instead, the 3GPP has defined the requirements in Release 4, pretty much finalized Release 5, and recently began working on Release 6, all of which are subsequent releases of the UMTS Release ’99 standard.
Further adding to the confusion, within each release (e.g., Release ’99) there are multiple versions. For example, Release ’99 began with the March 2000 version of the Release ’99 standard, and has since evolved every three months since that time in conjunction with the quarterly 3GPP plenary meetings. Although the basic functionality of Release ’99 does not change each quarter, the technical definition of how the functionality is implemented does change. Specifically, 3GPP members submit change requests (CRs), which identify changes to the baseline documentation. CRs can include anything from typographical or grammatical errors to additional/changed text that is inserted/replaced to clear up an ambiguity or correct an error, both of which could prevent a successful launch, in the documentation.
Before we present some of our findings, we would like to point out a few points worth considering. Food for thought, if you will:
* Each subsequent version (the date of the first version is March 2000) of Release ’99 is most likely not compatible with earlier versions of the release, this is especially true for the earliest versions. For example, if one OEM uses the June 2001 version and another OEM implements the December 2000 version, it is likely that their solutions will not be compatible with each other. Recall, Nokia issued a press release in December 2001 stating that it had successfully completed a WCDMA call using the June 2001 version of Release ’99, following a successful WCDMA call using the December 2000 version back in August 2001.
* Until the Release is stable, it is difficult and impractical to introduce 3G handsets into the market. In particular, software patches and upgrades are required in both the base stations and in each vendor’s handsets for each new version. Although at some point, 3G devices should be forward compatible with newer versions, thus relinquishing the need for a software upgrade in the handset, this may not be the case for initial fixes that are required, or especially if a vendor misinterprets the standard.
* Since the RAN, Core and Terminal Working Groups work separately from each other, each group needs to ensure that changes it makes to its respective documentation do not impact work being done by other Working Groups. Although the 3GPP and 3GPP2 strive for tight configuration control, mistakes can occur and an impact to another group may not be identified and corrected until after the fact.
* Once the standard is relatively stable, test scripts are required to test the infrastructure, the handsets, and the interoperability of the handsets and infrastructure for all of the vendor’s solutions against the most recent version of the standard. Complete and accurate IOT (interoperability testing) cannot begin until after the standard is stable and after the test scripts have been developed. Further, it can literally take years, not months, to successfully complete the initial set of tests, especially when problems are identified, subsequently fixed and then re-tested, not to mention developing additional test scripts and subsequent testing/fixing to address all of the little “quirks” that are noticed during network trials.
Release ’99 Status.
Now that we have identified some points worth considering, we would like to draw your attention to the results of our research. Our view is that some vendors in the industry have suggested that all work on the Release ’99 standard is complete, since the standard is “frozen.” However, the term “frozen” only means that the 3GPP has decided what functionality to include or exclude from the release. Once the functionality is “frozen,” the technical aspects of implementing that standard remain. In presenting this information, we are not trying to discredit the tremendous of amount of work that has been accomplished to date. Instead, we are merely trying to point out that a lot of work remains to be done before Release ’99 is ready for widespread commercial launches (we’ll get to Release 4 and Release 5 later in this report).
The 3GPP last met in early March and our findings are based upon draft meeting notes, the 3GPP work plan and discussions with industry contacts who have insight into Release ‘99.
Test Scripts Still Need to Be Developed.
The 3GPP decided to use the March 2002 version of the core specifications to develop test scripts for the 3G terminals. In the past, once a certain version has been identified, it has taken seven months to develop the test scripts. Changes to the March 2002 version will also have to be included in the test scripts, although it is believed that future changes will have minimal impact to IOT. Still, this implies to us that it will optimistically be mid September before IOT can even begin; obviously, it will be later before commercial networks will be ready.
Change Requests Continue.
Since the beginning of the year, approximately 300 Release ’99 change requests have been submitted to the 3GPP, and most were adopted (nearly 500 CRs were submitted in the last three months of 2001). Many of the CRs involved changes to the RAN or the terminals, including test procedures. Although some of the CRs appear to us to be inconsequential (typos), a meaningful percentage of the CRs involved corrections or clarifications to the existing standard.
Base Station Solutions Vary by Vendor.
Two separate industry contacts, that have good insight into the major vendor’s base station solutions, informed us that the test equipment they are providing to each vendor to test their base stations require different parameters in the test equipment. The implication is that each vendor is interpreting the standard differently or testing their base station to a different version of Release ’99.
Waiting for Release 4.
We have heard rumblings within the industry that, given continued Release ’99 delays, some carriers are considering waiting until Release 4 to launch their 3G service. It has even been suggested by some contacts in the industry that Release ’99 handsets may not be fully compatible with Release 4 infrastructure. As a result, once the carriers upgrade their RAN and Core to Release 4, the already distributed Release ’99 3G handsets may not work properly, if at all.
So what happens?
Our belief, based upon these data points, is that there will not be any successful commercial launches of Release ’99 in 2002. Those carriers that do launch commercial service will most likely utilize a single vendor for its terminals and infrastructure. For example, Sonera plans to launch 3G on September 26, 2002. Nokia is providing the handsets and the network infrastructure.
Other operators will use an earlier version of Release ’99 (e.g., June 2001 instead of March 2002), which buys them some time to conduct IOT, but which does not provide interoperability with other networks that may be using a later version of the release. We are aware of at least one network in Europe where the operator is using the June 2001 version of Release ’99, even though the 3GPP is advocating that terminals are tested against the March 2002 version of Release ’99.
Given strong suggestions about forward compatibility issues, it is also foreseeable that carriers who launch 3G service “early” (we use the term loosely) will do a “soft” launch. By that, we mean that the carriers will distribute a limited number of handsets, perhaps even single-mode WCDMA phones, on a trial basis to pre-selected trial subscribers. Once the carrier has sufficiently tested the network, the carrier will ask subscribers to exchange their handsets for a new handset that is fully compliant (and works in other networks).
Another important consideration for successful early 3G launches is the handoff between the UMTS and GSM/GPRS networks and vice versa. This capability is especially important since initial UMTS networks will have very limited geographical coverage, thus depending upon GPRS to provide a majority of the coverage. In our view, handoff problems will continue to loom well beyond 2002 due to the technical challenges of handing off between different technologies, not to mention the work remaining to develop a complete set of test scripts to ensure vendor and network interoperability.
FOMA is a hybrid of Release ’99, but it will evolve to become compatible with the UMTS standard. Before we move on to Release 4 and Release 5 of the UMTS standard, we need to clear up any misunderstanding you may have about NTT DoCoMo’s FOMA service and its relationship with Release ’99. NTT DoCoMo is a member of 3GPP and it is still involved the 3GPP process. However, it elected early on to deploy 3G service before the Release ’99 standard was frozen. Its FOMA service, therefore, is based upon a prerelease version of the Release ’99 standard. Since DoCoMo went at it alone, its 3G solution has evolved, but has since evolved and is not compatible with Release ‘99. However, DoCoMo has committed to making its FOMA service compatible with the UMTS standard in the next two years.
Release 4 Update.
The 3GPP is also working to finalize Release 4. Like Release ’99, the functionality in Release 4 is “frozen” (in March 2001), but the change request process continues.
In the first three months of 2002, members of the 3GPP submitted nearly 1,400 change requests pertaining to Release 4. However, the magnitude of CRs should drop each quarter. Further, Release 4 activity is somewhat inconsequential, since most carriers—at least the early movers—are still focused on launching Release ’99.
Release 5 and Beyond Update.
The 3GPP had scheduled the Release 5 standard to be “frozen” in March 2002. However, June 2002 is now the target date for the release to be frozen, since some functionality requires additional work. For example, work on HSDPA (High Speed Downlink Packet Access) appears relatively on track to be done in June. However, other functionality, like IP Multimedia Services, has work remaining.
At this point, Release 5 is probably a 2005 deployment scenario, at best. Even if the release is “frozen” this summer, it will still undergo a lengthy CR process. However, what concerns us most is that major OEMs are proposing vastly different network architecture solutions to implement the same release. As a result, vendors are distributing the functionality of Release 5 across different, and often newly introduced, hardware components. In some cases, our initial impression is that carriers will have to scrap some Release ’99 hardware and replace it with completely new hardware, especially if they elect to switch vendors. Recall, the intent and desire of the operators is to have a seamless migration to the next release—the “Upgradable” in “All the Right STUFF.”
Since our last report, the 3GPP has begun to define Release 6. Although we do not know a lot about this release, early indications suggest that it will include MIMO (multiple input, multiple output) antenna technology and other enhancements to the radio network, including WLAN-UMTS interworking.
END
==========
[[ This is me, now. ]]
Earlier this evening, I posted Release 99 corrections (with Release 4 and 5 reflectors) from (2) 3GPP Working Groups for only the month of August. Meetings occur monthly.
It is my impression that corrections - that denote changes to the standard - have not, and are not, slowing down.
Are the latest test scripts for 3G terminals that were based on the March 2002 version of the core specifications of any value today? Or are they antiquated?
Deutsche's lucid explanation of 3GPP's development process details the disastrous ramifications of continued corrections to any UMTSwCDMA development schedule.
Further, Deutsche sheds light on the exceptional opportunity that Nokia had to design, develop, manufacture and test Sonera's 3G network - to select its preferred Release 99 standard version and provide all components... infrastructure, handsets and software - with none of the concerns of standard version or vendor incompatibility that will be typical for very nearly all 3GSM carrier networks.
Nokia had almost the latitude that DoCoMo had in developing FOMA - except FOMA development was not chained to Release 99 outline specifications, and includes proprietary NTT IP. In light of Nokia's results with Sonera, how anxious is NTT to "upgrade" its networks to bona fide Release 99?
Indeed, only in this context can one grasp the gravity and magnitude of Nokia's failure to deliver a viable, commercial network to Sonera on schedule.
When will a stable Release 99 standard exist?
I wouldn't care to guess.
Will it even matter? FOMA suggests that it may not... DoCoMo is flying free of the standard - and all the incompatibility issues - and still can't make it fly.
What are 3GSM carrier launch schedules worth in the interim?
Squat.
And as Clark noted earlier today, it's not just the handsets... it's not just the infrastructure... it's the standard. |
