Private-network operators
Turn network observations into a clearer view of what went wrong, which actions are appropriate and when an engineer should intervene.
RESEARCH-LED NETWORK INTELLIGENCE
Intelligent networks.
Trustworthy decisions.
MAVERIC is developing secure AI control for private 5G and Open RAN. We connect network evidence, decision checks and recovery—with operators in control.
RESEARCH & DEVELOPMENT Rooted in research at RPTU · Kaiserslautern, Germany
MAVERIC / RESEARCH VISION01—03
INTELLIGENCE WITH OVERSIGHT
Observe. Understand. Act safely.
01
Security before execution.
02
Recovery with evidence.
03
Learning with local data.
01 / WHO WE WORK WITHCONVERSATIONS ACROSS EUROPE
Complex networks.
Shared operational challenges.
When equipment, applications and control policies interact, finding a fault is only the beginning. The next challenge is deciding what to change—and knowing that the change helped.
Telecom systems integrators
Connect evidence across vendors and define how new intelligence fits into existing management systems and operational workflows.
Industrial connectivity teams
Translate service interruptions into testable scenarios, so new automation is evaluated against the needs of your operation.
Have a challenge in mind? Let’s explore it together.
02 / THE CONCEPTTRUST AT THE POINT OF DECISION
Three research pillars.
One coordinated approach.
AI can help operate complex networks. But an incorrect action, a policy conflict or unreliable data can put live services at risk. MAVERIC brings security, recovery and learning into one coordinated research concept.
SECURE AI CONTROL
Verify before
the network acts.
Research threat detection and checks on proposed control actions, so safety constraints and operator policies can be considered before a change reaches the network.
AUTONOMOUS SELF-HEALING
Understand the cause.
Guide the recovery.
Investigate policy conflicts and network instability. Develop recovery actions that respect operator intent, support rollback and explain what changed.
FEDERATED INTELLIGENCE
Learn across sites.
Keep raw data local.
Research shared model improvement without centralising operational data, with checks for model drift, unreliable updates and safe deployment.
These capabilities are in development. MAVERIC’s current goal is an integrated research demonstrator.
03 / OUR APPROACHOPEN INTERFACES. COORDINATED CONTROL.
More automation.
With operator control.
Our proposed architecture connects network observations, decision checks and recovery. The aim is to make every step traceable, from detecting a problem to verifying the outcome of an action.
We’re designing for gradual automation: recommendations first, operator approval where needed, and autonomous actions within defined boundaries.
Explore the technical architecture
MAVERIC stands for Modular AI-driven Virtualized Enhancement for RAN Intelligence and Control.
The design combines management and orchestration (SMO), longer-term decisions in the non-real-time RAN Intelligent Controller (Non-RT RIC), and near-real-time control (Near-RT RIC). Modular applications, called rApps and xApps, coordinate through open O-RAN interfaces.
- STI: Security & Threat Intelligence
- APH: Autonomous Policy & Healing
- FMIE: Federated Model Intelligence & Evolution
PROPOSED O-RAN INTEGRATION
Manage the lifecycle
Deployment, versioning, audit records and recovery authority.
RIC
Understand and coordinate
Policy generation, diagnosis, model training and aggregation.
RIC
Validate and act
Live telemetry, bounded-latency inference and safety checks on control actions.
Open interfaces · Modular rApps and xApps · Operator-defined policies
FROM OBSERVATION TO VERIFIED OUTCOME
Close the loop.
Keep the human in control.
A proposed workflow for the demonstrator. Each action should have supporting evidence, a defined authority and a way to check the result.
- 01 / OBSERVE
Collect evidence
Bring together available telemetry, events and network context.
Network observations - 02 / UNDERSTAND
Assess the cause
Investigate anomalies and competing explanations for a fault.
Diagnosis & supporting evidence - 03 / CHECK
Review the action
Check operator policies, constraints and approval requirements.
Bounded recovery proposal - 04 / VERIFY
Evaluate the outcome
Compare results with the intended outcome; escalate or roll back where needed.
Outcome & audit record
Operator oversight across the loopDefined permissions · Approval boundaries · Traceable decisions
04 / WHERE IT MATTERSPRIVATE 5G & INDUSTRIAL CAMPUS NETWORKS
When connectivity
keeps work moving.
Illustrative settings for future validation—not customer deployments. Each requires a clear service objective and access to relevant network observations.
Manufacturing
When connected equipment loses reliable service: investigate network anomalies before proposing changes that could affect production.
Logistics
When mobile robots experience inconsistent connectivity: examine mobility, performance and competing control policies together.
Healthcare
When campus services depend on connectivity: explore resilient operation and collaborative learning that keeps raw operational data local.
05 / RESEARCH INTO PRACTICE
A clear path to
a tested proof of principle.
Our existing foundation is a machine-learning detection application evaluated on a laboratory Open RAN testbed for one family of denial-of-service attacks. This establishes a starting point for research; it does not yet validate the complete MAVERIC system.
The integrated demonstrator is in development. Our goal is to evaluate secure control, recovery and learning together in a realistic environment with equipment from multiple vendors.
- 01
EXISTING FOUNDATION
Laboratory validation
Security detection on a research testbed.
- 02
DEVELOPMENT FOCUS
Integrated demonstrator
Self-healing and federated intelligence remain at the concept and early experimental stage.
- 03
RESEARCH GOAL
Realistic validation
A proof of principle tested in a relevant environment, targeting Technology Readiness Level 5.
Research behind the direction
Selected work co-authored by Prudhvi Kakani on Open RAN security. These publications provide background to our approach; they are not validation of the integrated demonstrator.
XAPP SECURITY · 2024
Evaluation of xApps and their Security in Next-Generation Open Radio Access Networks
Research into application vulnerabilities, threats and security approaches in Open RAN.
E2 INTERFACE SECURITY · 2024
Exploring the Implications and Methodologies of Securing the E2 Interface
Research into security threats and protection approaches for the interface between the controller and radio network.
06 / BEHIND MAVERICRESEARCH. ENGINEERING. APPLICATION.
Research expertise.
Practical ambition.
MAVERIC connects Open RAN security research with the practical questions of operating complex networks. Based in Kaiserslautern, we’re building relationships with operators, integrators and researchers across Europe.
Prudhvi Kakani
Founder & Product Lead
Manjunath Chavva
Technical Lead
QUESTIONS, ANSWERED
A clear starting point.
What to know before we talk.
Can I buy or deploy MAVERIC today?
MAVERIC is a research-led startup initiative. We are developing an integrated demonstrator, rather than offering a production-ready platform. Conversations today focus on operational requirements, integration and potential validation scenarios.
Who should get in touch?
Private-network operators, telecom systems integrators, industrial connectivity teams and research collaborators. A specific recurring fault, difficult diagnosis or concern about automated changes is a useful starting point.
Does MAVERIC replace our network management tools?
Our proposed approach is to complement existing management and orchestration systems with coordinated security, diagnosis and decision checks. Compatibility will depend on the interfaces, telemetry and control permissions available in each environment.
Will AI make changes without an operator?
Our design starts with recommendations and operator oversight. Any future autonomous actions would need explicit boundaries, policy checks and a way to verify the result. The level of automation is part of the research and validation work.
What would an initial collaboration involve?
We would first discuss your network environment and an operational challenge. Together, we can assess what observations are available, identify a bounded test scenario and agree how a useful result would be measured. No sensitive network data is needed for the first conversation.
07 / COLLABORATIONA EUROPEAN PERSPECTIVE
Let’s start with
your network challenge.
Are you operating a private network, integrating equipment, or exploring trustworthy automation? We’d like to hear what is difficult today and what a useful solution would need to do.
Our ambition is to support European technological sovereignty through open interfaces, operator control and research that can transfer into real deployments.
Useful starting points include time spent diagnosing a fault, unnecessary site visits, or uncertainty about an automated change. Together, we can define what a meaningful improvement would look like.
What we can explore together
- Share an operational challengeDescribe a recurring fault, a manual task or a concern about automated network decisions.
- Explore integrationDiscuss your network stack, available data, interfaces and deployment constraints.
- Shape a future pilotIdentify a test scenario, success criteria and a potential evaluation setting as the demonstrator develops.
Email Prudhvi
Tell us your organisation, network environment and the challenge you’d like to discuss.
Trust at the heart
of network autonomy.