One way in which 5G promises to be very different from its predecessors is in its focus on open platforms and interfaces above the physical layer. Several high profile events in the past weeks have shone the spotlight on the essential requirement for standard interfaces throughout the 5G network, and this work is increasingly being driven by open initiatives as well as by conventional standards bodies.
The Facebook-initiated Telecom Infra Project (TIP) and Linux Foundation-hosted ORAN Alliance are both working on common ways to interface the remote radio head, in a disaggregated RAN, with the virtualized baseband. There were strong hints, at the recent TIP Summit in London, that the two might converge their efforts, which are largely complementary, driving a de facto standard which could run with any fronthaul technology and many of the 3GPP’s eight defined splits between virtualized and physical network functions.
However, a great deal of work on standard interfaces has been pioneered in the enterprise or small cell worlds. Small Cell Forum has worked on open interfaces between different elements of the network since its formation over a decade ago.
In the macro network, operators pushed for a common fronthaul connection between a baseband unit (BBU) and a remote radio unit (RRU). They ended up with CPRI, an over-complicated and semi-proprietary specification dominated by a few vendors.
At the same time, the Forum came up with Iuh, an interface between small cells and controllers or cores, and succeeded in having it adopted as a 3GPP standard. It went on to enhance the TR-069 and X2 interfaces, working with other standards bodies; and then to create FAPI and nFAPI. The next step will be to take these into the denser, more open world of 5G.
FAPI (functional application platform interface) provides a set of common APIs to support interoperability between the 3G, 4G or 5G PHY, and software elements such as the security coprocessor or scheduler. nFAPI (network FAPI) extends the concept to virtualized small cell networks and provides an interface between the remote radio unit (RRU), and a centralized baseband unit (BBU), on which some or all of the baseband functions are virtualized.
These two sets of APIs map to the two main approaches to dense 5G deployment. In outdoor HetNets, 3GPP is already defining some of the interfaces to connect RRUs with local or far-away BBUs, but FAPI will still be important as an internal interface between the chipset and software layers (a gap in the 3GPP or open source standards efforts). In indoor enterprise networks, nFAPI will be essential to support the most common architecture – a group of cells controlled by a central, virtualized unit.
In other words, SCF is addressing a requirement that is absolutely central to the economics of 5G, since these, for many operators, rest on the ability to disaggregate the RAN. To date, deployment of this virtualized RAN (vRAN) has been impeded partly by the lack of a fully unified fronthaul interface between the RRU and BBU. SCF has already made considerable progress on this, in a small cell context, and is also addressing aspects of the interface which which are not a central focus for other initiatives, filling gaps in work by 3GPP or the open projects.
Densification absolutely requires an open ecosystem to make deployment of large numbers of cells cost-effective. These interfaces are the enabler of the innovation and price competition that comes when operators can select equipment from many suppliers.
As outlined above, the move towards open interfaces has been more advanced in the small cell layer than the macro network. Several suppliers already offer architectures in which a number of small cells are clustered around a centralized, virtualized controller. Once standard interfaces between the radio and the controller are supported, along with open baseband virtual network functions (VNFs), the economic argument for densification will be far stronger, and a key disadvantage vis-à-vis WiFi (its open ecosystem) will be removed.
nFAPI took a big step towards larger scale when an open initiative, the open-nfapi project, was announced to support it in February. Led by Cisco, the project has worked to implement the nFAPI disaggregated base station architecture in a set of libraries and simulators. As Mark Grayson, distinguished consulting engineer at Cisco, wrote in a blog post, Cisco was drawn to nFAPI because it supports active RAN sharing, and therefore neutral host or multi-operator small cell networks – which will be essential for many enterprise and vertical industry deployments in the 5G era, which cannot be tied to a single MNO.
“We concluded that the Small Cell Forum’s multivendor nFAPI split architecture, together with its neutral host management model, offered a new approach to active sharing of an LTE RAN based on a multivendor CU-DU implementation,” wrote Grayson.
He highlighted the challenges facing many would-be open standards, particularly in getting sufficient real world adoption when there are many open, and even open source, projects to choose from. He continued: “Even though Cisco has licensed the nFAPI libraries under the permissive Apache 2.0 licence – enabling them to be integrated by the widest possible set of stakeholders, including closed source proprietary RAN products – we do not underestimate the real challenges in getting such capabilities implemented.”
To help address the issue, Cisco joined the Open Air Interface (OAI) open source initiative and said it would use its membership to demonstrate how to integrate the open-nFAPI libraries into an existing LTE RAN protocol stack; the integration between the lower nFAPI libraries and the PHY layer, implemented on a software defined radio platform; and the integration between the upper nFAPI libraries and the MAC and RRC layers. This will help to lower the barriers to active RAN sharing in small cell environments.
Such efforts show how specifications like nFAPI will certainly be working in a friendlier environment in the 5G era, with the whole industry shifting towards a more open focus, and with enterprise interest in 5G driving operators towards a more neutral host approach.
The risk is that there will be too many competing interfaces, and work that was supposed to end fragmentation will actually contribute to it. Most groups acknowledge that they will need to find common ground, and in the case of the mooted TIP/ORAN cooperation, that might be possible because the groups have very different focuses. TIP works primarily on product specs and the physical layer, and ORAN defines the higher layers of a virtualized architecture.
Two critical success factors for open RANs were highlighted by that conversation – the need for cooperation between projects, and the need for this work to enable practical deployment in the near term.
5G, to be successful, needs to enable a far greater variety of services and network behaviors than 4G, and that will affect the deployment. A machine-to-machine service requiring ubiquitous coverage of a city, for instance, will require cells to be positioned and coordinated in a very different way from an enterprise application requiring a limited zone of very high capacity. The key will be uniform specifications for equipment and interfaces, which will enable limitless flexibility in how the cells are deployed and which form factors and suppliers are included.
All this highlights the important role projects like nFAPI can play in accelerating the progress towards a fully deployable, commercially viable dense 5G network. This will be critical to the small cell industry and help smooth the path to 5G densification. However, as small cells become a central element of all RANs, it will also be important for the influential new open RAN groups to get even closer to SCF, enriching their own work by tapping into SCF’s interfaces.
The OpenAirInterface Software Alliance:
The mission of the OpenAirInterface Software Alliance (OSA), established in 2014, is to provide software and tools for 5G R&D and product development.
It states its aims as bringing open source software, running on general purposes processors such as x86 or ARM, deep into the RAN, an area whose demanding requirements have led to continuing dominance by “proprietary elements that stifle innovation and increase the cost for the operators to deploy new services/applications”, as the OSA puts it.
With software-defined networking (SDN) and open source technologies starting to change the landscape in other parts of the network, and in devices, the group believes an open source implementation of a real time stack (base station, terminal and core) on general purpose processors is needed.
The OSA provides a framework for intellectual property and financial contributions to the platform. It provides a standards-compliant implementation of a subset of 3GPP Release 10 LTE for UE, eNB, MME, HSS, SGw and PGw on Linux-based computing (x86 or ARM architectures). The software is freely distributed by the Alliance under the terms of the OSA license model.