When a group led by AT&T pushed a fast-track 5G New Radio standard, Non-Standalone (NSA), through the 3GPP, it enabled operators to deploy a 5G RAN more quickly and easily than they had expected – but just succeeded in storing up problems for the future.
Because 5G NR NSA uses the LTE core, most operators which have launched 5G already have been able to make relatively simple upgrades to their RANs – in most cases, deploying very limited numbers of base stations, using existing site infrastructure, and often simply being able to activate a previously installed 5G-ready base station. This has some benefits – enabling faster speeds and some new applications in congested urban areas; supporting fixed wireless access (FWA) for MNOs in markets where this is appropriate; allowing the MNO to make the all-important 5G marketing claims.
But as we have often analyzed, the NSA network severely limits the impact of 5G, because it makes it hard to support multivendor interoperability – a key element of MNO’s economic targets for 5G – and it does not include the 5G core.
As operators wanting to replace Huawei as a RAN supplier have found, it is difficult to do this when the 4G base stations, and the new 5G ones, come from different vendors. The situation is even more challenging in a disaggregated RAN where multiple radio units share a separated centralized baseband unit.
And true 5G can only be achieved with the 5G core. The core will be essential to support the really transformative promises of 5G, such as agile service delivery, full multilayer convergence with wireline and other wireless networks, critical communications and network slicing. When AT&T proposed NSA, opponents – such as Orange, Telefónica and even Nokia – argued that 5G meant little in terms of network transformation unless the RAN and core were deployed together. A quick fix in the RAN would just delay MNOs in deploying the core and so the real benefits – for the operators themselves, and for consumers and industries – would be delayed too.
At the 5G Core Summit conference in Madrid this week, Telefónica was reiterating those concerns, while outlining just how challenging the deployment of true 5G would be. Migrating to that core will be a significant and complex process, especially as many operators will not be satisfied with existing, but cumbersome, virtualised architectures based on OpenStack and virtual machines.
Instead, they will take the opportunity of the core replacement to introduce more flexible, cost-efficient cloud-native architectures – virtual network functions for the core, and eventually the RAN, which were designed from scratch to run on cloud infrastructure, in the data center or at the edge; and which make use of containers and microservices to make it quicker to design, deploy and reconfigure hundreds of different services.
All this means a lot of pain is stored up for the operators over the coming few years, in terms of deploying the core, and ensuring their cores and RANs can contain elements from multiple suppliers. A few will start on the path in 2020, starting to introduce the 5G core and the accompanying 5G NR Standalone (SA) RAN. China Mobile is one – the only MNO which originally said it would go straight to NR SA, but balked at the task and is deploying NSA. However, it plans to start implementing SA and network slicing from next year.
Many will wait for several years before embarking on the 5G core, perhaps waiting for cloud-native technologies and standards to stabilize. Even those which move quickly towards the new core generally have far longer timelines to virtualize their RANs – at least beyond small cell and in-building clusters. A few operators, such as Telecom Italia, are making strides towards large-scale vRAN but for most, this will be the final frontier of fully fledged 5G.
What may make the progress towards multivendor, cloud-native, fully software-defined and programmable 5G networks simpler, cheaper and quicker? Some are pinning their hopes on open network initiatives such as Telecom Infra Project (TIP) and Open RAN (ORAN) Alliance to accelerate the pace. Certainly, ORAN is showing signs of moving on from being an interesting open source architecture to something deployable. Last week, it announced its first testing and verification programs (see Wireless Watch September 20 edition), and this week, Japan’s NTT Docomo has described a multivendor, ORAN-based network which it plans to deploy as a pre-commercial 5G system ahead of next summer’s Tokyo Olympic Games (see below).