The ANDIAMO suite in brief
The ANDIAMO Suite is a software package developed by Orange Innovation to address operational automation needs related to optical and microwave transport networks. It includes several modules:
- ANDIAMO Data collects data to build a Data Lake encompassing all optical and microwave network information for Orange.
- ANDIAMO Network Design facilitates network design and engineering.
- Fiber RootCause AI Ready, which I contribute to, uses AI for diagnosing optical and microwave network issues.
Use of AI in Developing This Solution
The idea stems from extensive reflection and active listening to Orange subsidiaries' needs, along with insights from international scientific conferences. This assessment identified multiple use cases, notably network fault diagnosis, prioritized due to its importance for technicians' daily activities.
Given the network's constant evolution—new services, equipment, and software updates—AI is well-suited for network diagnosis, thanks to its ability to learn, adapt, and evolve alongside the network.
Challenges encountered
Since project initiation, several challenges arose:
- Data collection was complex, requiring convincing operational teams to access network data and test the solution in real conditions, achieved with Orange France.
- Integrating various suppliers and developing data collectors was essential for comprehensive optical transport network coverage.
- Raw, unlabeled data required technician involvement for labeling, a process still ongoing.
- Internal resistance was addressed by reassuring technicians about AI's benefits, alleviating concerns about job impact.
- Developing dual expertise in optical transport networks and AI was necessary for project success.
- Managing interactions among research contributors and delivery teams was crucial to maintain motivation, facilitate knowledge transfer, and ensure project continuity and growth.
Benefits and key lessons
The developed solution uniquely considers the specificities of optical layers, ensuring optimal AI adaptation. A patent was filed for clustering phases, incorporating proven natural language processing techniques tailored to alarm and network diagnosis contexts. This involves multi-step processing: topology-based clustering, temporal analysis, and semantic validation of alarm clusters.
Another notable feature is multi-vendor compatibility, enabling integration with all Orange France equipment, setting it apart from existing solutions. The goal is to extend this approach to IP and microwave networks for a comprehensive network view.
Currently, the solution is in production at Orange France, managing approximately 50 critical incidents daily, and is accessible on Orange Poland’s pre-production network. Technician feedback indicates a time saving of about 30 minutes per incident, translating to 25 hours saved daily. This efficiency allows technicians to better anticipate diagnostics and reduces reactive, urgent interventions.
Future developments
Planned enhancements include automating certain fault resolutions, such as restarting a card, and optimizing technician interventions through incident geolocation. Additional equipment vendors will be integrated as needed, and automatic technician support will be strengthened.
Collaborations are ongoing, with increasing interest from entities like OINIS. Orange aims to reach a 4-level automation on the TM Forum scale, currently at 2.5. An RFP has been issued to improve operational efficiency across all network areas. The ANDIAMO Suite acts as a facilitating module, integrating key data for the chosen solutions and paving the way for broader adoption within subsidiaries.
This solution exemplifies the synergy between AI, optics, and software deployment, reflecting Orange France and Poland’s commitment to innovation. Close collaboration with their teams has enabled field testing and current achievements.
