German telecom giant Deutsche Telekom continues to tackle complex integration challenges of disaggregated 5G network infrastructure, with its latest moves around network slicing and the radio access network (RAN) intelligent controller (RIC). Both involved multiple vendors, highlighting the potential benefits and complexity of a disaggregated network architecture.

The network slicing work was a proof of concept (PoC) that showed the capability to automate the instantiation of an on-demand network slice with guaranteed quality of service (QoS) running on a 5G standalone (SA) network ordered by an enterprise through a customer portal.

Deutsche Telekom provided the business support system (BSS)-infused customer portal, Blue Planet provided the multi-domain service orchestrator (MDSO), Mavenir supplied the cloud-native 5G core, and Ericsson the RAN and network management. All of those elements were connected through TM Forum supplied open APIs.

That concoction was able to support the placement of an order for an end-to-end QoS network slice; the MDSO then taking that order, deconstructing its components and distributing the service requests to the appropriate vendor; the orchestration of the RAN and 5G core to support the service; the service itself instantiated as a network function; and the enterprise end-user devices subscribed and provisioned for the service. Claimed time to completion: “minutes.”

A Deutsche Telekom executive noted the importance of the PoC toward supporting enterprise (read: paying) customers.

“Dynamic slicing and automation are key innovations that will play a crucial role to enable the delivery of on-demand 5G offerings with guaranteed quality of service for our enterprise customers,” Changsoon Choi, VP of network service differentiation and convergence at Deutsche Telekom, wrote in a statement.

Deutsche Telekom earlier this year worked with Ericsson on creating a secure 5G network slice that tied directly into a private cloud. That PoC also used TM Forum-based APIs to integrate third-party management interfaces that allowed external management systems to incorporate slice ordering and management.

Analyst firms have for years been touting the financial benefits of commercial network-slicing capabilities, especially tied to penetrating different market verticals.

ABI Research noted that it still expects a dynamic market over time, but that current technical and use cases issues have slowed those forecasts. The firm had initially forecast a $66 billion market opportunity by 2026; however it’s now forecasting a $19.5 billion market by 2028.

“Horizontal integration for cross-domain interoperability is critical going forward. Equally important is vertical integration for 5G slicing lifecycle management of multi-vendor deployments,” Don Alusha, senior analyst at ABI Research, wrote on the ongoing market challenges.

“There is ongoing market activity for the 5G core network penetration and maturity of 5G slice management functions. To that end, enterprises will seek to create and reserve slices statically and on demand. They also want to efficiently integrate with cloud providers through open and programmable [APIs] to enable hybrid cloud/cellular slice adoption. [Network equipment vendors] and other suppliers offer solutions enabling [communication service providers] to create fully automated and programmatic slicing capability over access, transport and core network domains.”

Deutsche Telekom’s latest network slicing cut followed a carrier-led multivendor trial of a programmable RAN that used non-real-time RIC and rApps to automate and optimize a disaggregated RAN. That work was then put into a lengthy white paper highlighting continued challenges and necessary areas of focus.

The work was done with VMware, Juniper Networks, Viavi Solutions and AirHop. Those firms used ONAP and O-RAN specifications to onboard and deploy closed-loop rApps on a non-real-time RIC to support a pair of use cases. rApps are microservice-based applications operating in non-real time that provide an operator with more control over their open RAN environments.

The first was the optimization of physical cell identifier (PCI) to detect and resolve “PCI confusion and collision scenarios.” The second used an artificial intelligence (AI) and machine learning (ML) model to train and determine the optimum time to enable or disable “sleep-mode” on a cell site to balance power requirements with capacity need.

The potential energy savings has been a key component of RIC work as operators look for ways to manage power usage across 5G network equipment from different vendors. An ABI Research report indicated that while 5G equipment can be more power efficient than legacy 4G LTE equipment, network densification will increase the overall power draw.

“Despite 5G consuming less power than 4G per unit of traffic, the overall energy consumption is still much higher, driven by more power-thirsty radios and network densification,” ABI Research analyst Fei Liu wrote in the report. Liu added that whereas 5G infrastructure can be 90% more energy efficient than 4G LTE on a per bit basis, a typical 5G base station also requires three-times more energy to provide the same coverage as a 4G LTE network.

“5G energy consumption depends on radio configuration, hardware and traffic load, and over 70% of the consumed energy is in the [RAN]. A 5G RAN consumes up to 2.7 kilowatts of power with [64-transmit/64-receive] massive [multiple-input/multiple-output] configurations in a typical condition, whereas an LTE radio consumes about 0.8 kilowatts,” she wrote.

Petr Ledl, VP and head of Deutsche Telekom’s network trials and integration lab, said in a statement that the operator would take results from the PoC “to accelerate non-RT RIC/rApp development toward production readiness.”