Cloudly and the LF Connectivity community are excited to announce the release of Maveric v1.0—now available on GitHub: 🔗 Download Maveric v1.0.0
This release marks a major milestone: Maveric now includes all the components needed to function as a true RAN Digital Twin, enabling proactive, AI-powered optimization for Mobility Robustness Optimization (MRO), Coverage and Capacity (CCO), and Energy Savings.
Why This Release Matters
Modern mobile networks are too complex for static tuning. Operators need dynamic, data-driven systems that simulate, predict, and optimize network behavior under real-world constraints—without disrupting live service.
Maveric v1.0 delivers just that.
Whether you’re solving dropped calls, minimizing interference, or reducing power consumption, Maveric now has the full stack of simulation, analytics, and AI-ready modules needed to support digital twin–driven operations.
What’s New in v1.0
This version consolidates months of development across multiple areas of the RAN lifecycle:
- SINR_DB Helper Function Fix
Ensures accurate signal quality modeling and stability across simulation runs
By @WaterMenon09 in PR #25 - Traffic Demand Generation Suite
Models realistic user traffic profiles across time and geography to support CCO and energy optimization
By @tanzim10 in PR #20 - Energy Saving App
Optimize energy consumption with Maveric’s modular Energy App, which uses reinforcement learning and a Bayesian Digital Twin to intelligently control cell activity and antenna tilt based on time-aware traffic patterns—ensuring power savings without compromising service quality.
By @Lkishor123 in PR #24 - Load Balance App
Dynamically balance network load with Maveric’s modular RL-powered Load Balancer, which adjusts antenna tilts based on temporal traffic patterns using a Bayesian Digital Twin for fast, offline simulations—optimizing coverage, capacity, and user experience.
By @tanzim10 in PR #22
Use Cases Supported in v1.0
Maveric now supports simulation and optimization for the following key domains:
Mobility Robustness Optimization (MRO)
Test and tune handover strategies using both heuristic methods and reinforcement learning. Fine-tune critical parameters such as hysteresis and time-to-trigger (TTT) to minimize dropped calls, ping-pong handovers, and radio link failures all within a safe, high-fidelity simulation environment.
Coverage and Capacity Optimization (CCO)
Simulate signal quality, interference, and user density across varied environments using real-time telemetry and high-resolution terrain data. Optimize antenna tilt, azimuth, and transmit power to minimize dead zones and interference. Evaluate configuration impacts through both real and synthetic scenarios for precise, data-driven tuning.
Energy Savings
Simulate off-peak traffic to identify power-saving opportunities using a modular Energy App powered by reinforcement learning. The system dynamically turns cell sectors on/off and adjusts antenna tilts based on time-of-day traffic patterns, balancing energy efficiency with QoS (Quality of services). It leverages a Bayesian Digital Twin RF model for fast, localized simulations, enabling efficient multi-day training cycles decoupled from backend delays.
Load Balancing
Balance network traffic dynamically using a modular reinforcement learning pipeline designed for Coverage and Capacity Optimization (CCO). This application adjusts cell antenna tilts hourly based on multi-day traffic patterns to optimize load distribution, coverage, and service quality. It uses a pre-trained Bayesian Digital Twin RF model for fast local simulations, ensuring efficient training without relying on live backend systems.
Traffic Load Generation
Simulate realistic multi-day UE traffic demand across a configurable mobile network using Maveric’s Traffic Load Simulation Framework. By leveraging Voronoi tessellation from cell tower coordinates, the framework dynamically allocates UEs based on spatial-temporal weights, capturing movement patterns like residential-to-commercial transitions. This synthetic dataset supports network analysis, ML model training, and coverage optimization.
Technical Capabilities
- RIC-compatible modules for real-world deployment
- Jupyter notebooks and test harnesses for rapid prototyping
- API documentation for easy integration
- Support for scenario-based training and inference with synthetic or real-world data
- Robust data generation suite for on-demand data synthesization based on adjustable scenarios
- CLIs provide streamlined access to core Maveric simulation, training, and deployment functions enabling control of digital twin workflows without manual intervention.
Why It Matters to the Community
With Maveric v1.0, we’ve evolved from focused mobility optimization through our MRO engine to a comprehensive AI-native Digital Twin platform purpose-built to simulate, learn from, and intelligently optimize radio access networks across coverage, capacity, mobility, and energy dimensions.Operators and researchers now have the tools to:
- Evaluate “what if” scenarios before rollout
- Compare policies for handovers, load balancing, and energy use
- Move toward closed-loop RAN automation
This is open-source infrastructure that helps you operate like the most advanced carriers without the cost and complexity.
Get Started
Dive into the code, run the examples, and start optimizing:
đź“– Maveric GitHub README
What’s Next
We’re actively collaborating with partners to integrate Maveric with real-world data sources — including mobile apps, field surveys, and live RAN telemetry — to ground simulations in operational reality. Looking ahead, planned enhancements include:
- 5G and Wi-Fi integration to support multi-access network optimization across both licensed and unlicensed spectrums
- Web-based GUI for interactive “what-if” network planning, including visual telemetry overlays and optimization suggestions
- Deeper integration with AI model training pipelines, enabling adaptive policy generation from both real and synthetic data
Let us know how you’re using Maveric—or how we can help tailor it to your deployment.
👉 Contact Cloudly | Contribute on GitHub