rApps Collaboration for Measurable RAN Improvements

Marcin Hoffmann

2026-02-24

Mobile Networks, Research, Wireless Technologies

Currently, RAN automation is identified as one of the crucial targets for 5G and 6G networks. This is expected to be achieved with the main focus on the Service Management and Orchestration (SMO)/Non Real-Time RAN Intelligent Controller (Non-RT RIC) hosting dedicated applications named rApps. These can target energy savings: Cell On/Off Switching rApp (COOS-rApp), or Load Balancing (LB): Traffic Steering rApp (TS-rApp). However, before these solutions could be deployed in RAN, a reliable verification is needed, which requires real-world data and Network Digital Twin – NDT (see Digital Twin – What Is It and How Can It Affect Future Networks).

From this perspective, in this blogpost we are excited to share a glimpse of our recent collaboration with VIAVI Solutions on rApps operation under live-network data. In detail, we show how we integrated our NDT solution with Viavi’s Geolocation rApp, exposing the location-dependent datasets based on real-world network KPI. Having that in place, we put two of our key rApps to the test using real-world network data: COOS-rApp and TS-rApp.

Data flow description

*Fig. 1. Illustration of the data flow between Geolocation rApp, NDT, and TS/COOS rApps*

While from the architectural point of view, both NDT, TS-rApp, COOS-rApp, and Geolocation rApp reside in SMO/Non-RT RIC, here we are focusing on how the data is exchanged between them. This is demonstrated in Figure 1. Basically, the VIAVI Geolocation rApp is capturing the raw network data and performing internal analytics to transform them into location-aware datasets containing, e.g.,

Estimated user location,
User radio channel properties,
User connection duration,
User data volume.

This location-dependent data, together with RAN topology (e.g., obtained through the O1 interface), is then fed into the NDT. This approach enables recreating real-network conditions, and therefore tests rApps in a safe environment to assess their potential abilities to enhance RAN performance.

Rimedo Labs Network Digital Twin

The NDT, which consumes location-dependent data from the VIAVI Geolocation rApp, is a tool developed by Rimedo Labs to develop AI/ML-based RAN automation algorithms as rApps. One of its key features is the ability to create synthetic datasets based on the real-world data from live networks (see our blog post Synthetic Data Generation for AI-based O-RAN Apps).

*Fig. 2. Overview of the Rimedo Labs NDT*

The high-level overview of the Rimedo Labs NDT is presented in Fig. 2. In general, NDT is a system-level simulator of the O-RAN network. It has a modular architecture enabling the flexible definition of evaluation scenarios under various parameter sets. In particular, the Rimedo Labs NDT supports:

API for training of AI/ML-based RAN algorithms like xApps/rApps and testing their mutual influence (see O-RAN Virtual Exhibition MWC 2025 Barcelona)
Synthetic data generation for offline AI/ML training
Collecting logs and KPI visualization
3GPP-compliant modules including: Antenna Models (TR 36.942, TR 38.803), Power consumption (TR 38.864), Massive MIMO (TR 38.921), or Propagation models (TR 38.901).
Predefined cell deployments (e.g., regular grid) along with real-world cell layouts loaded from external data
UE deployment, both predefined (e.g., random deployment) and based on real-world data (e.g., trace-driven or imported from measurements)

Results

During our collaboration with VIAVI, we integrated the VIAVI Geolocation rApp with our NDT described in the previous section. The dataset covered the area of about 38 km², with 135 cells deployed on 16 base stations of 10/20 MHz bandwidth and the carrier frequencies of 800, 900, 1800, and 2100 MHz, respectively. Under this setup, our NDT recreated the 40-minute-long traffic pattern generated by a number of users oscillating between 500 and 800. This pattern was repeated for 3 scenarios:

#1 Default – without rApps involved (marked with a blue line),
#2 TS-rApp – with TS-rApp operating (marked with a yellow line),
#3 COOS-rApp + TS-rApp – with cooperating TS-rApp and COOS-rApp (marked with a green line).

*Fig. 3. Network outage (% of users achieving less than 95% of requested throughput)*

Under this scenario, we first evaluated network outage (% of users achieving less than 95% of requested throughput), which is visible in Figure 2. We can see that utilization of TS-rApp allows to reduce outage by about 50%. This means that high-quality service is provided to more users compared with a Default approach.

What is also important is that the operation of COOS-rApp does not introduce a negative effect on network outage. This means that some cells can be switched off in this situation to reduce power consumption. This is presented in Figure 3. Throughout the simulation, the average power consumed can be reduced from about 89.4 kW to about 80.39 kW, giving about 11% of power savings.

Conclusions

A crucial part of the rApp development lifecycle is the verification of the multiple cooperating applications against the real-world data.

A key enabler for this is to leverage the Geolocation rApp from VIAVI Solutions, which Rimedo Labs successfully integrated into their NDT. The obtained geolocation insights allowed us to recreate a large-scale traffic pattern covering hundreds of cells and users, and test two Rimedo Labs solutions:

Rimedo Labs TS-rApp – Delivered approximately 50% improvement in QoS fulfillment,
Rimedo Labs COOS-rApp – On average, achieved 11% reduction in power consumption through intelligent dynamic cell switching.

These results demonstrate what’s possible when the advanced Network Digital Twin meets real operational data.

See more on these topics in the following Rimedo blogs:

VIAVI Geolocation technology resources

Author Bio

Marcin Hoffmann is a Technical Solution Manager at Rimedo Labs, working on O-RAN software development solutions and R&D projects covering energy savings, traffic steering, and massive MIMO. Marcin is a Graduate Student Member at IEEE and received the M.Sc. degree (Hons.) in electronics and telecommunication from Poznań University of Technology in 2019, where he is currently pursuing a Ph.D. degree with the Institute of Radiocommunications. He was involved in many both national and international research projects. His research interests include the utilization of machine learning and location-dependent information for network management. He coauthored many scientific articles published in top journals like IEEE Journal on Selected Areas in Communications, IEEE Transactions on Intelligent Transportation Systems, IEEE Communications Magazine, or IEEE Access.