Sprint has teamed up with Ericsson to develop a virtualised core network and operating system to support its IoT services. This is one of the most ambitious elements so far in the MNO’s strategy to move to a virtualized, white box network, to transform its costs and provide agile support for a wide variety of services.
The core, which will launch along with 5G from 2019, is specifically optimized for IoT applications which require very low latency and/or critical availability, as well as high security. It ticks many of the hot boxes for the expansion of operators’ revenue streams via low latency 5G/IoT, such as edge compute, artificial intelligence (AI) and robotics.
It remains to be seen how Sprint leverages those hot enablers for commercial gain – and how its merger with T-Mobile USA, if approved, will alter its architectures. Sprint has always been prepared to be advanced, even daring, in its technology choices, but has a far weaker track record in translating those investments into commercial advantage.
The USA’s fourth operator has not been in the forefront of the push towards software-defined networking (SDN) and agile platforms, but over the past year it has been more positive about the potential for a transformed architecture for 5G, and before he left the firm in November, former COO Gunther Ottendorfer explained in a company blog how Sprint was using an NFV OpenStack cloud platform to support development of its LTE Advanced and Gigabit LTE networks and prepare for 5G.
Sprint has also been deploying common infrastructure data centers across its network and has a roadmap to expand them, following a “cap and grow” approach to commercializing NFV. This means capping any expansion of legacy core network hardware while adding new functionality and capacity on the virtualized platform.
Most importantly, Sprint has been building a new virtual core that gradually replacing single-purpose, bare metal platforms with a unified NFV infrastructure (NFVi), running its packet core and IMS capabilities as virtual network functions (VNFs). The new IoT core will be based on this architecture, and though details are sparse (more will be shared next week at the MWC Americas show), they suggest that Sprint is taking a more advanced approach than many operators to one of the key issues associated with the mobile IoT – the deployment of edge compute.
Edge-based processing and storage will be essential to some low latency IoT applications, reducing the time spent transferring data to the cloud for analysis. Many operators believe there is a logical alignment between their physical locations, such as cell sites and central offices, and edge nodes.
However, that view – enshrined in ETSI’s original Multi-access Edge Compute (MEC) architectures – is very limiting. It constrains operators to applications which are, indeed, well-served by an edge that is relatively far from most end users. These tend to be the consumer edge services, such as video caching, which the MNOs know well, but which are challenging when it comes to boosting ARPU.
For many IoT services, a far more distributed edge, extending into enterprise premises, and even homes or devices, will be required. A few operators are looking to deploy edge nodes with small cells to address the need for far greater edge coverage, but they are rare, and their hesitation may see neutral host, cloud or industrial service providers leaping into the gap.
In Sprint’s case, though, the virtualized IoT core is designed to be highly dispersed with “nodes distributed right to the enterprise premise, if necessary, to support specific security, privacy, and latency requirements,” said Sprint and Ericsson in their joint statement. And because it is fully virtualized, it can be easily extended to additional edge nodes – which do not necessarily need to be deployed by Sprint. By divorcing the core and applications from the physical location, the operator has greater flexibility to extend its edge cost-effectively while retaining a strong place in the value chain, moving above mere connectivity and compute.
The core design “reduces distance between the device generating the data and the IoT application processing it,” the partners continued. That enables the operating system, which accompanies the core, to support combined connectivity management, device management, data management, and managed services while “delivering immediate intelligence” on the data collected.
“Configuration and updates of firmware and software are managed for each device. All data is managed securely with world-class security on the chip level. The IoT OS provides full subscription lifecycle management and monitoring of billing and usage data,” the companies said.
The managed services element includes service assurance for IoT devices and locations, such as “network operations centre monitoring, service resource fulfilment, cloud orchestration management, and application management”.
And the virtualized nature of the system means it is not tied to 5G – there will be support for all generations of cellular machine-to-machine (M2M) connectivity including Cat-M and NB-IoT, within the 4G standards. Sprint is to support Cat-M later this year while T-Mobile, its future spouse, has an NB-IoT development.
Sprint’s SVP of IoT, Ivo Rook, said: “We are combining our IoT strategy with Ericsson’s expertise to build a platform primed for the most demanding applications like AI, edge computing, robotics, autonomous vehicles and more.”
The new offering was devised with the help of Ericsson’s IoT Accelerator, which was first launched in April 2016. This packaged up a combination of Ericsson products and services, with an integrated marketplace for partners’ hardware and services. Ericsson said it would target larger industrial and public contracts, helping customers overcome the cost and complexity barriers of trying to build their own platform from scratch to support all the elements of an IoT roll-out.
“We believe in empowering organizations to connect people, devices, and things as quickly as possible. We will initially focus on public safety, utilities, transport, and smart cities customers, continuously adjusting to their needs and feedback, and improving the ease of use and delivery speed through a DevOps approach to software developments,” said Orvar Hurtig, Ericsson’s head of industry and society, on unveiling the Accelerator.
The Swedish vendor’s Device Connection Platform and Ericsson Cloud System are key underpinnings of the offering, though customer and hybrid clouds are supported too. The Cellular for Massive IoT software is also an enabler, promising to support and manage millions of connections per base station over low power networks.
As part of the service, Ericsson provides initial set-up, systems integration and consultancy. The marketplace partners provide a repository of applications, tailored to specific geographic regions, and supported by regional Ericsson offices.