by Simon Thompson
Network slicing is the great hope for a brand new business model for 5G, one that would encourage a wealth of new connected services and revenue streams. Telefonica has been a trailblazer, especially in its cooperations with Huawei, but its latest slicing claims are based on work with a start-up, Netsia. It has demonstrated an end-to-end slice which includes the RAN (the toughest part to slice, especially before 5G) – this could support new service models, and also shows how operators believe the new software-defined telco network will open the doors to a new supply chain full of innovative players and price competition.
In network slicing, a virtualized, programmable network would allocate the right combination of resources to support optimize connectivity for a particular application or user, for as long as they needed it (from a long term contract to a few minutes). The orchestrator or slicing engine would call up the necessary elements (fixed and mobile, virtual and physical, optical and IP, storage and processing power – with the right type of radio connection, such as low latency for a critical communications application).
The ultimate vision is that the service or user themselves would automatically call up the virtualized resources they required, with no intervention from the operator. The operator might be a telco, a vendor like Nokia or a webscale player – running this automated, highly flexible network from the cloud and monetizing it through added value services, and through access to differentiated connectivity and other resources.
Telefonica has contributed to this vision with its own R&D and its contributions to group efforts like the European Union-backed 5G PPP’s NORMA project. Its User Centric No Cell (UCNC) work with Huawei, which has been submitted to 3GPP as part of the 5G standards process, describes how a user (which could be a ‘thing’) would dial up the resources they needed at any one time, and instead of logging onto a certain cell, the mobile signal would effectively follow them around.
Now the Spanish group claims it has fitted “the last piece of the puzzle” for end-to-end network slicing by integrating a programmable software defined network (SDN) into a RAN.
It has used technology called ProgRAN, developed by US start-up Netsia, to do this. Netsia virtualizes key functions of the base station – wireless channel resources and radio management modules – while leaving just the physical elements at the site.
Telefonica used the Netsia technology in conjunction with an virtualized LTE packet core at its labs in Madrid, testing a hospital use case in which the RAN was shared while also being optimized for use by different groups with different connectivity requirements (doctors, patients, visitors and connected objects all had their own slices with different network characteristics).
It said the ability to program allocation of resources right down to radio level – most so-called slicing platforms today stop at the fixed network – is a breakthrough. Juan Carlos García, Telefonica’s technology and architecture global director, said in a statement: “The integration of Netsia’s Virtual LTE RAN Platform into a commercial LTE core in our Labs is an important step forward to demonstrate that an end-to-end network slicing for 5G networks is possible, from the core to the RAN, and how it can be applied to specific service environments, like that of a virtual private LTE network.”
Oguz Oktay, Netsia’s VP of wireless solutions, added: “ProgRAN’s dynamic RAN slicing capability carries the potential to link network services to new commercial opportunities for operators. ProgRAN allows the RAN to programmatically adapt itself to different service requirements and customer experience needs. This means that an operator will be able to offer network functions to many different industries, such as IoT, healthcare or automotive, using a RAN-as-a-service business model.”
These statements show that Telefonica and Netsia are initially focusing mainly on localized virtualized RANs or subnets, which are a priority for many of the operators investigating vRAN. This is partly because they can be deployed discretely, allowing the MNO to test vRAN capabilities without risk to its main RAN and core; and partly because these subnets – often deployed using small cells and a local, virtualized packet core – can enable new revenue streams in high value markets. For instance, an enterprise might have its own subnet to support business critical services, or a city might install one to support high reliability, low latency services like traffic management.
These users would gain control of these important activities, while paying the MNO a premium for enabling the system – provided, of course, that these subnets remain in MNO spectrum. When they are deployed in unlicensed or shared bands such as 5 GHz or the US’s CBRS band in 3.5 GHz, the MNO loses its control, especially if technologies like MulteFire are used (the emerging MulteFire supports LTE in 5 GHz or 3.5 GHz without requiring a host network in licensed spectrum, unlike the current cellular options for unlicensed bands, LTE-U and LTE-LAA). The disruptive potential of unlicensed cellular subnets has not been lost on MNOs like Telefonica – which want to snap up the market first using technologies they control – or on cablecos, which believe this could be a route into the premium wireless market for them (see separate item on cable).
Of course, true slicing promises to go well beyond support for long term subnets or static slices. These are enabled using current technologies including small cell clusters with local controllers like those from Spidercloud; virtualized local packet cores like those from Quortus or Core Network Dynamics; flexible cloud-based platforms like Nokia’s Cloud Packet Core; and even end-to-end IP VPNs (virtual private networks) with mobile access. True 5G slicing will support very flexible, on-demand, potentially ephemeral connections and services, which will be paid for on a per-usage basis, probably using a marketplace system.
This will be one of the chief ways to expand the business case for a large MNO like Telefonica, and to justify its hefty investments both in virtualization and SDN, and in 5G. The operator’s huge Unica virtualization platform is due to go live in several markets, including Argentina and Colombia, by the end of the year. Eventually it will be deployed in 13 European and Latin American markets.
It includes a wide range of vendors, large and small, as Telefonica – like AT&T with its Domain 2.0 program – seeks to shake up its supply chain at the same time as it shakes up its network and its cost base. Ericsson has overall integration responsibilities while Huawei – a key R&D partner with Telefonica in the area of slicing – is supplying the virtualized EPC (evolved packet core). This will be based on Huawei’s CloudEPC platform and will support LTE voice and data service delivery in Spain, Germany, Brazil, Argentina, Uruguay, Mexico, Colombia, Peru, Panama, Costa Rica, Nicaragua, El Salvador and Guatemala.
Telefonica says its vEPC will not just make delivery of conventional voice and data services more flexible and efficient, but will help it to deploy new business cases, such as large-scale MVNO support, IoT services, private LTE networks and applications enabled by MEC (Multi-access Edge Computing).
Late last year, Telefonica also brought ZTE into the Unica fold by commissioning it to build a virtual IMS (IP Multimedia Subsystem) for several Latin American markets; and it is using Nokia virtualized IP routers in Spain. However, Ericsson is also the lead overall vendor for Unica roll-outs in Germany and Argentina.
This year’s Mobile World Congress was the fourth in a row at which Telefonica offered major showcases of Unica, which spans the core, transport and access networks plus IT systems and CPE. The first commercial Unica infrastructure deployment was in Germany last year, based on Ericsson’s new Hyperscale Datacenter System 8000, and the vendor’s Cloud Manager, Cloud SDN, OpenStack-based Cloud Execution Environment, plus automation and systems integration.
One of its cooperations with Huawei is in UCNC, which gives every device a permanent identifier, placing the responsibility of delivering a good experience onto the RAN. It aims to deliver more reliable and efficient services to users, compared to current mobility management techniques. In their lab trials, the partners say they have increased the number of connections per cell by 233%, reduced signalling overhead by 78% and decreased latency by 95%, compared with currently commercial LTE networks.