What leads to a successful 5G sourcing project?

Having worked with infrastructure sourcing for over 20 years, Northstream has inevitably came across 5G in the last couple of years. Northstream’s well proven sourcing methodology seems to still hold up well in the 5G era, driving the process with strong momentum and transparency which translates to short lead times and fact-based decisions.

The foundation to a successful RFP is usually laid early in the process when we (Northstream and the client together) take a holistic approach to define the 5G launch/rollout strategy and key evaluation criteria for vendor proposals. While each operator’s unique position could lead to different answers, the questions we asked have a lot in common. And we would like to share our thoughts on some of the most representative ones in this blogpost.

The pace of 5G rollout

  • This is by far the most critical question. We have long argued that the launch of 5G, at least in the first few years, will be used for boosting MBB capacity. But its effectiveness depends on not only the availability of network and spectrum but also that of the devices.
  • The NSA version of 5G smoothens the introduction 5G in the network but it drives up the device cost and complexity due to the need of an extra transceiver for simultaneous 4G and 5G transmission. This means that until the SA version becomes available, 5G devices would probably be limited to a few more pricy/premium options. In contrast to the rapid 4G uptake which was largely contributed by the switching from feature phone to smartphone, the handset replacement cycle has been slowing in recent years. A rapid 5G uptake based on not only MNO pushing devices but also consumer pull would require ample supply and variety of 5G devices from low to high end segments of the market.
  • Therefore, the decision of 5G rollout pace or the construction of the RFP scenario requires a careful balance between the CAPEX budget, network capacity need and a realistic outlook of the device uptake, which in turn requires a deep dive into the local market situation (competition and customer profiles)

What about new use cases and revenue streams

  • FWA is the most tangible one, although its size of opportunity varies from market to market. FWA requires mMIMO to deliver the necessary throughput to be competitive against cable or even fiber. But the current price point for mMIMO equipment dictates a very selective deployment, usually on top of existing infrastructure with good backbone connectivity, and covering a sufficient number of households, with not too many trees in-between…
  • Given the competition and the lock-down nature of fixed networks, there is clearly an urge to go to the market asap with FWA solution to capture this window of opportunity. But it takes time to identify the “sweet spot” for FWA, which is rather small compared to the typical mobile network rollout. Therefore, it is crucial to engage the business unit as early as possible to get a good feeling of where and how many FWA solution are needed. We have come across several cases where the cost of sales is underestimated, FWA is not a conventional retail product
  • Beyond FWA, we have unfortunately yet to identify any other concrete new use case. While there are many interesting and promising trials going on the the world it is still so that URLLC and mIoT are not yet standardized and no real commercial products from vendors either. The key here is to agree on a pricing model that limits the exposure to any “surprise” price tag for those future feature releases. An all-inclusive package can be very desirable from the operator’s point of view, but some flexibility should be allowed to be fair to the vendors who are also exposed to quite substantial uncertainties at the moment.

The choice of RAN architecture

  • Despite all the new capabilities of NR and the two-staged introduction of NSA and SA 5G, the build out of a 5G RAN is, at large, business as usual. In most parts of the world we have three major vendors to choose from. In our opinion Samsung has the technology but still lacks the global organization to expand its market share to have a sustainable business.
  • The blueprints for a 5G sites seem to be quite well aligned among the different vendor offers: one multiband radio for all low bands and one for all mid bands, all connected to the a single multiband antenna if space and wind load would allow it, then probably a separate mMIMO AAU for 3.5GHz band if needed; some variations exist in the baseband design approaches, some with better flexibility and some with better capacity, but a full-fledged 5G site might require multiple boards with one dedicated for providing the necessary 5G capacity, and another for legacy technologies (2/3/4G).
  • Perhaps the biggest decision point here is the choice of mMIMO AAU, whether to go all the way with 64TR or be more moderate, e.g., with 16TR. A higher order MIMO could bring significant benefits, but it is highly dependent on the surrounding environment and user distributions, at the cost of not only higher price but much higher energy consumption. The latter in particular is becoming a bane to the increasingly OPEX sensitive operators. Yet there is still no methodology concise and specific enough to allow a fair estimation/comparison of the energy consumption of different vendors’ offers.
  • C-RAN has always been discussed but rarely materialized in any large-scale deployment in our engagements. Its application has not gone beyond the basic baseband “hotel” in the absence of any concrete new use case such as MEC, or any real need for ultra-dense network with mmWave small cells yet. Without such concentration of baseband processing unit, the benefit of V-RAN is also limited (the most expensive component in RAN, i.e., mMIMO AAUs cannot be virtualized anyway)

The evolution to 5G core

  • Since 5G SA core would be fully cloud native, virtualization will be a must and no longer an option. Although NSA core can still be supported by an updated EPC (which is not necessarily virtualized), it is generally considered wiser to start the virtualization early to learn the ropes while the traffic is still limited. This also means core network functions are now shopped as “pieces of SW” allowing pick and mix solutions from different vendors. It is possible to choose the best of breeds in every area, but this comes at the cost of increased complexity, particularly in managing the multi-vendor integration and share of responsibility. The choice of the orchestration tool would be essential to ensure a smooth operation of such a complex setup.
  • The most likely winner in this virtualization process are the vendors with the best cloud-nativeness, which leads to a much smoother integration towards the NFVI and other VNFs, as well as oftentimes the best performance. Compared to RAN where there are very few vendors to select among the 5G core market will be much more competitive with new players entering the market.
  • But not all these differences are readily apparent in numbers stated in product specifications. Only trials and PoC can reveal the full complications and differences. Therefore, we recommend our customers to start PoC in the virtualized environment as early as possible, rather than relying too much on the evaluation of SoC answers…

These are of course just the tip of the iceberg, then there are also OSS, transmission issues to be dealt with. Let us know if you like to discuss more on the important aspects and complexities to address in 5G sourcing!


Lei is Senior Manager and Head of Technology at Northstream

Next read

"Give something to get something"

Smart city is no longer just a topic about the future. Many ‘smart’ / digital services are already operational today, ranging from public transport logistics, parking surveillance, traffic management, waste control, etc. They are, however, mostly deployed on top of existing communication backbone (mobile and fixed networks), built primarily for serving residential and B2B legacy communication needs. [...]
Read more