Cloudly is releasing comprehensive Mobility Robustness Optimization (MRO) tools in Maveric v0.2 that network engineers and researchers can use to eliminate dropped calls and optimize handovers in 5G networks.
Why This Release Matters
Every dropped call represents user frustration, and network optimization shouldn’t be guesswork. Mobile network operators struggle with handover failures and Radio Link Failures that degrade user experience, while traditional optimization approaches take months to deploy and often fail to adapt to changing network conditions. Our MRO tools solve these problems by providing data-driven optimization that works in real-world scenarios.
What We’ve Built
Realistic Network Testing Without Risk: Our UE mobility simulation framework uses digital twins of actual cellular sites to model user movement and network behavior. Engineers can safely test optimization strategies without disrupting live networks, while digital twins predict signal variations and connection patterns with production-level accuracy.
Clear Performance Insights: We developed MRO metrics that turn complex network data into actionable intelligence. By analyzing historical handover success rates, Radio Link Failures, and mobility events, teams get quantitative insights into exactly where their networks need improvement.
Two Proven Optimization Approaches: Our baseline heuristic framework jump-starts handover tuning by running a targeted grid search over both hysteresis and Time-to-Trigger (TTT) settings. In practice, this means simulating a wide range of margin and delay combinations—observing where small margins or brief triggers cause unnecessary cell “chatter,” and where overly conservative values lead to late handovers and dropped connections. By evaluating each pair across representative traffic and mobility scenarios, we quickly identify a robust, off-the-shelf configuration that balances stability and responsiveness.
For those who demand ever-greater precision, our Reinforcement Learning engine continuously refines these parameters in production. It watches live performance—tracking handover success rates, ping-pong handovers, and radio-link failures—and treats each event as feedback. When a certain hysteresis‐TTT combination yields smoother transitions, the model reinforces that choice; when conditions shift—say, new interference sources emerge or user speeds change—the RL agent adapts, nudging margins and timers toward fresh optima. The result is a handover policy that not only starts strong but learns and evolves alongside your network.
Production-Ready Implementation: Every component includes comprehensive unit tests and detailed Jupyter notebooks with real-world examples. The documentation and API interfaces are designed for immediate integration into existing network optimization workflows.
Technical Components
- Advanced Mobility Simulation
Accurately models realistic user movement and behavior across dense urban, rural, and high-speed scenarios using digital twins enabling safe, controlled testing of handover strategies without touching the live network. - Customizable MRO Metrics
Includes a comprehensive set of tunable metrics like handover success rate, radio link failure rate, and ping-pong events designed to support diverse optimization goals across different network types and conditions. - Dual Optimization Engines
Offers both a fast-start heuristic grid search for stable handover policy tuning and a reinforcement learning engine that adapts to live network conditions in real time for continuous performance improvement. - RIC-Compatible and Deployment-Ready
Built as a seamless plug-in to Open RAN ecosystems, Maveric integrates easily with RIC platforms, supports standard interfaces, and includes APIs and test coverage for straightforward operational deployment.
Testing and Documentation
- Complete unit test coverage for all MRO functionality
- Comprehensive Jupyter notebooks with step-by-step workflows
- Real network scenario examples and performance benchmarks
- API documentation for easy integration
Impact for the Community
These tools transform network optimization from reactive problem-solving to proactive performance management. Network operators can deploy optimizations in days rather than months, while researchers have access to proven algorithms and realistic testing environments. By open-sourcing our MRO implementation, we’re enabling the broader community to build better, more reliable mobile networks that adapt intelligently to user needs. For more, see the release note here.
To get started: https://github.com/lf-connectivity/maveric/blob/main/README.md
What’s Coming Next
Our next release will focus on Energy Savings optimization, a critical challenge as networks seek to reduce power consumption while maintaining service quality. We’re developing sophisticated traffic modeling capabilities that will demonstrate measurable energy efficiency improvements in realistic network scenarios. This work will provide the community with comprehensive tools to balance performance optimization with sustainability goals, addressing one of the industry’s most pressing concerns.