Indirect vs. direct data forwarding: A look at the debate
As mobile networks around the world are poised to make the switch from 4G to 5G, the question of how a smooth and efficient handover procedure can be delivered has been a key discussion driver across 3GPP. We take a look at both sides of the debate below.
In the 3GPP community, we all want the same thing: to find the best possible technological solutions to drive progress in the world. This does not mean, however, that all telecom entities, operators, and working groups always agree on the finer details of our collective way forward.
One such example of this occurred in the RAN3 meetings held in the Autumn of 2018, in which we at Ericsson occupied an unusual position as the champion of a rather unpopular opinion. The issue in question? How best to forward buffered user plane data from a source to its target node in a 4G to 5G handover procedure.
Debating the best approach for 4G to 5G handover
A handover denotes the process in which a radio access network (RAN) changes its radio transmitters, radio access mode, or radio system to allow the transmission of information as efficient as possible. Simple enough. To illustrate, we encounter this handover process when a mobile phone user moves within their service provider’s network, switching between cellular towers as they move in and out of range of the network towers, as well as when a user’s device must switch technologies (from 4G to 5G, for example) or even between operators themselves.
As one of the key requirements of the 5G system, data forwarding at the point of the handover helps to minimize jitter, data loss, and out-of-order delivery – understandably important features. In all handovers, solutions are implemented to execute the process quickly and smoothly – qualities upon which both current and upcoming 3GPP discussions will focus to bring further improvement.
It was on this topic of data forwarding that a debate arose between whether the solution should follow:
- the quality of service (QoS) and bearer model between a 5G core network and 5G radio access network (NG-RAN)
- or the much simpler model typically applied between NG-RAN nodes for data forwarding, proving its applicability for data forwarding between RAN nodes located in different systems
The former requires a ‘translation’ between models, whereas the latter would bypass this translation process altogether, skipping the intermediary step and going directly to the relevant RAN node. Summed up, the limelight was on direct versus indirect data forwarding.
Direct vs. indirect data forwarding
Proponents of indirect forwarding argue in favor of its centralized management, level of control, and ease of deployment and configuration for new devices. Ericsson, however, took the alternative view during the RAN3 meetings, highlighting the unnecessary complexity involved in including more system entities than needed (as indirect forwarding dictates). It is self-evident that minimizing impacted entities when selecting a solution – especially in multi-vendor deployments – is crucial, as unnecessary complications will inevitably go on to hinder and prolong any potential future enhancements to the solution.
Here at Ericsson, we proposed that direct data forwarding on the other hand, is an easy-to-understand, simplified solution, requiring the fewest possible user plane functions. This entails a lower impact on both RAN and core networks and a more direct (hence the name) approach to quickly and easily transferring data. Translated, that’s increased efficiency and reduced complexity. In our view, just as today, the case for direct data forwarding seems to be far stronger in comparison.
Accepting a compromise for the benefit of consensus
This was the main contention we stood behind in the fall of 2018, when we were able to effectively prove the technical feasibility of direct data forwarding as a standard solution (for a technical overview, download the document R3-186804 from 3GPP).
It wasn’t until RAN3 meeting #102 that the solution’s feasibility was finally acknowledged (after hours of both online and offline discussions) by all members of the working group. Unfortunately, though, by this point there was no stopping the momentum behind Release 15, and it was not possible to select the solution that had the least impact on the overall system. Therefore, in the end, we made the decision to compromise, establishing a working agreement and accepting Release 15’s support of indirect data forwarding to conclude the specification with an intermediate conversion step.
With this being said, we want to emphasize that we do not challenge the implementation of Rel-15 or the method for indirect data forwarding for inter-system handovers. While we do not find justification for the technical reasons laid out in the RAN3 discussion, we still of course respect the views of our peers in 3GPP.
And, for us, this is what 3GPP is all about: finding a majority consensus for the best possible solution in that moment and moving the discussion forward. We accept, of course, that we will not always agree with the opinion of the majority.
Looking ahead to future releases
Not all was lost though, and we were able to specify in Release 16 an option for data forwarding at inter-system handovers (which we believe to be a superior solution compared to the alternative proposal). As always in 3GPP, this was only possible thanks to compromise, whereby, in this case, our option would be applicable only if connectivity for direct data forwarding existed. In June of 2019, RAN3 finalized its specification work, with the Release 16 specifications still to be created in June 2020.
In summary, the great value of the solution is a simplified implementation. The entire functionality is handled in the RAN, and no support from the core is required. In addition, the need for signaling is significantly reduced. To be a bit more detailed:
- The data passes directly between the nodes eNB and gNB. No bearer conversion is needed.
- Direct data forwarding will be used as part of the NR intra-system handover solution, and can be re-used for data forwarding between 4G and 5G.
- The core network user plane function is used only to relay data between 4G and 5G. Since there is no other impact, complexity is reduced.
Although our originally proposed solution was not ultimately chosen in the RAN3 meetings, we were able to make a powerful case for our belief in breaking down complexity and bringing simplified solutions to the industry. In the end, a compromise was made that will hopefully pave the way for a more effective 5G system, with the future of direct data forwarding retaining a competitive potential for implementation.
About Ericsson and 3GPP standardization
3GPP is the body responsible for global standardization of 2G, 3G, 4G and 5G mobile networks. As a technology leader with an extensive patent portfolio, Ericsson places strategic importance in leading 3GPP initiatives that strive to ensure that the best possible solutions are included in new telecom standards. In our series of Standardization Stories, we showcase that, through leadership and innovation, Ericsson plays a leading role in building today's and tomorrow's global eco-system of Core and Radio Access Networks.
Want to find out more?
Gain a broader view on Ericsson’s role in shaping future telecom standards.
Read our previous 3GPP posts to find out how we have been setting new standards in: