Since the first building blocks of SON were laid down around 2008 by 3GPP and NGMN, uptake in SON deployments has been very selective by a few leading carriers for some use cases. However, universal applicability remains elusive. To say the least, the SON market is struggling – but why, and how that can be turned around is what interests me. Having just attended the SON USA conference, I had made a few observations and like to put some down here.
The context: SON building blocks were laid by NGMN and 3PGG in 2008 and have progressively been revised and updated to widen the scope of SON. They used a bottom up approach to define SON use cases for LTE which has expanded with every new 3GPP release of this technology. Specifications on 3G are more limited and follow from those of LTE. Applications of SON to the macro cell has been limited to a few use cases such as configuration and provisioning (neighbor relation is one of the most used features).
The operator perspective: There are multiple sentiments aired by operators when it comes to SON. There are questions on the value proposition which is difficult to quantify. For activities that can be streamlined, operators have developed in-house processes that substitute external SON systems. Operators are also more prone to test the water with the SON system provided by the RAN vendor rather than opt for a third party SON. With this approach, operators aim to limit investment in SON. This makes more sense wherever vendors are managing operator networks – especially in this case, SON becomes a feature of the RAN that the OEM can have a complete lock on. Network engineers perceive SON as a threat in the worst case. In the meantime, SON can be a contentious domain between different functional groups within the operator organization. Operators are highly vocal about having a multi-RAN SON system, yet this is ironic since a single SON system invests power into a single SON vendor.
The vendor perspective: The vendor space can be divided into RAN equipment vendors and third parties. RAN vendors have the advantage of easy access to data that the network elements generate (OEMs can easily hamper third parties’ access to this data). However, they don’t have monopoly on smarts and third party vendors differentiate by having innovative solutions that actually solve specific problems for operators. The third parties have specifically focused on 3G networks. Yet, some of the third party solutions have a narrow focus while some of the RAN OEM solutions struggle in terms of performance.
What’s next: escape forward! This sums up the state of SON. One emerging concept is pairing SON with big data analytics. While this is an interesting idea, the devil is in the details. Analytics target a certain use case – a well defined problem which is solved by customizing a process and algorithms. Coupling SON with data sciences requires good knowledge of both spaces. How the benefits are imparted to the network still remains to be seen especially as a closed-loop approach forms the basis of such pairing. Operator resistance to closed-loop processes limits the effectiveness of this new approach.
SON is widely viewed as essential for HetNets and while the uptake in small cells has lagged market expectations, it is not strange that SON has lagged correspondingly. But waiting for HetNets to take off means, to me, that it will be many years before SON sees some traction: The pain is not large enough yet to warrant its application.