QDXray

We’re proud to share that QDXray has received the award for Best AI-Driven Observability Solution.

https://www.eubusinessnews.com/winners/qdxray-bv/

This recognition reflects the work we’ve been doing to bring practical, operations-ready AI into broadband networks. QDXray focuses on turning large-scale DOCSIS and PON telemetry into clear insights, root-cause explanations, and automated actions that help operators prevent issues rather than simply react to them.

From real-time spectrum analysis to anomaly detection and autonomous workflows, our goal has always been the same: make complex networks easier to run and service quality easier to protect.

Thanks to our customers and partners who helped shape the platform and pushed us to solve real operational problems instead of theoretical ones. More to come.

What makes this especially powerful is when cable and fiber observability are no longer siloed.

Operators increasingly run hybrid access footprints:

• Legacy HFC
• New fiber overlays
• Migration zones
• Brownfield upgrades
• Rural fiber builds alongside urban DOCSIS networks
• Running separate tools for each technology fragments operations, data models, and workflows.

A unified observability platform changes that:

• One topology model across coax and fiber
• One analytics engine for RF and optical telemetry
• One outage detection framework
• One visualization layer
• One API surface into ticketing and workflow systems
• One operational truth for NOCs and field teams

Whether the impairment sits in a coax cascade or a fiber feeder — the operational process becomes consistent, automated, and data-driven.

Because passive networks don’t speak for themselves, observability becomes an analytics problem first — and a monitoring problem second.

Modern access observability relies on:

• Pattern recognition across thousands of alarms
• Behavioral baselining of optical levels and RF metrics
• Spatial clustering of failing endpoints
• Topology-aware correlation
• Time-based degradation detection
• Root-cause inference across shared infrastructure

AI-driven algorithms can:

• Distinguish mass outages from household faults
• Separate optical degradation from power failures
• Infer likely break points in feeder fibers or splitter trees
• Predict failing segments before customers call
• Suppress alarm storms while escalating real incidents

Instead of reacting to single-device alerts, operations teams gain a network-wide understanding of what is really happening in the field.

Traditional network management systems are great at answering questions like:

• Is my OLT port up?
• Is the CMTS interface flapping?
• Did this node generate an alarm?

But access networks demand different questions:

• Which passive segment is failing?
• Which group of homes shares the same damaged fiber?
• Is this degradation gradual or sudden?
• Is this outage correlated with excavation activity?
• Are we looking at optical degradation, RF impairment, or power loss in the field?

Those answers don’t live in a single device.

They emerge only when you correlate telemetry, alarms, topology models, OSS data, GIS layers, service inventories, and historical behavior patterns.
This is where observability begins.

Core and aggregation layers are full of routers, switches, and active probes producing rich telemetry.

The access network? Not so much.

In DOCSIS environments, huge parts of the topology consist of:

• Fiber nodes
• Amplifiers
• Taps
• Coax segments in the field

In FTTX networks, the challenge is similar but even more distributed:

• Feeder fibers and distribution cables
• ODFs and cabinets
• Splitters and splitter cascades
• Handholes and street-level closures
• Splice trays and field joints

These elements don’t emit telemetry.
They don’t send alarms.
They don’t speak SNMP.

Yet when something goes wrong — a degraded splice, a bent feeder fiber, water ingress, a construction crew hitting a duct — this is where the failure actually happens.

In the dynamic landscape of PON-based access networks, traditional network management and service assurance are no longer sufficient to ensure optimal performance and reliability. The advent of observability has revolutionized the way we understand and monitor these complex networks, providing invaluable insights and proactive capabilities. Let's dive deeper into the specific differences between plain network management, service assurance, and observability, and explore how observability is reshaping the industry.

🔍 Plain Network Management: Plain network management focuses on monitoring the network's hardware and infrastructure elements. It primarily aims to ensure the network's operational state, track connectivity, and maintain configuration consistency. While it provides basic visibility into the network, it falls short when it comes to understanding the system's behavior, performance bottlenecks, and root causes of issues. Plain network management tends to be reactive, primarily relying on fault management and reactive troubleshooting approaches.

🔒 Service Assurance: Service assurance takes a step further by monitoring and ensuring the quality and availability of services delivered over the network. It focuses on measuring service-level metrics, such as latency, packet loss, and availability, to guarantee that service level agreements (SLAs) are met. Service assurance primarily relies on predefined thresholds and performance indicators to trigger alerts and notifications. While it provides valuable insights into service performance, it often lacks the contextual understanding of the underlying infrastructure and its impact on service delivery.

🌟 Observability: Observability is a paradigm shift in network monitoring and management. It embraces a holistic approach that combines telemetry data, rich context, and advanced analytics to provide a comprehensive understanding of the network's behavior, performance, and health. Unlike traditional methods, observability empowers network operators with real-time insights, proactive anomaly detection, and the ability to uncover hidden dependencies and correlations within the network.

Most service providers today maintain a well-defined set of KPI’s, mostly based on telemetry from active network elements. This morning the NOC raised a P1 after seeing a few hundred PON ONT’s falling offline from PON ports on different line cards on different OLT’s.

The engineer on-call was alerted and quickly powered up his laptop to check what was wrong. Seeing drops all over the place he escalated to network planning to have a look into their fiber planning system and check topology on various endpoints.

Around 13:00, after hours of struggling through extensive logging, clicking through endless fiber planning screens, writing down data in excel sheets, it appeared a cable was damaged just next to a handhole.

You recognize this?
This is one of the pain areas we have to address when operating cable- and fiberplants. A data driven approach where data from all sources are combined in a data-science melting pot is the base for smart software algorithms creating actionable items driving corrective as well as proactive workflows in your organization.
This is QDXray, the battlefield is the access network based on a myriad of passives connected by complex fiber infrastructures.
One of our missions is to help our customers to minimize their day to day operational challenges and drive proactive workflows.
By the way: did you know QDXray offers WiFi additions to create the real end-to-end view? Just because QoE does not end at the frontdoor.

Data Integrity often is a challenge in Telecom operations. The more manual interventions are enabled and the less atomic provisioning workflows are constructed, the higher the probability of data pollution or data-integrity breaches in the provisioning chain.

Topology administration, the locations of passives, zipcode-housenumber to passive-infra, a misconfiguration lurks around every corner. Provisioning, new builds, or just an engineer making a weekend quick-fix, helping customers out quickly ‘first thing on Monday will be administering the quick change, but never happens..’.

By smart combining data from various datasources and applying a fair share of datascience algorithms, QDXray can support service providers in data cleaning projects and even provide data for automated workflows.