Purpose

Smarts is used to analyze and identify root causes of network problems in DSL and GPON (PON) last-mile access networks, where traditional monitoring methods are ineffective due to passive infrastructure and complex cabling structures.

Problem Overview

Last-mile networks consist of massive numbers of cables, splices, and distribution frames.
In theory, the topology forms a clean tree structure — but in reality, it’s entangled, with branches that reconnect or even service direction is sometimes reversed.
Challenges:
Difficult to locate the exact faulty cable segment.
Vendor tools lack correlation capabilities.
Passive optical cables cannot be directly monitored.
A single fault may impact users dispersed throughout the network.

Smarts Solution Principles

Since the passive network offers no direct monitoring, analytics rely on CPE (Customer Premises Equipment) data such as availability and performance.
Root cause analysis must cope with topology complexity, which makes rule-based or suppression-type correlation engines useless.
Smarts’ CodeBook correlation model is suitable because it defines all potential problems and their relationships (symptoms and causes) across topology objects.

Analytical Model in Smarts

Defines relationships between:
Client Circuits, Ports, Connection Points, Cable Segments, and Cables.
Layers and dependencies modeled as N:M and N:1 relationships.
Differentiates between:
Monitored objects (e.g., ports, client circuits).
Unmonitored objects (e.g., passive cables, splitters, distribution frames).
Uses inheritance and layered-over relationships to propagate fault states through the topology.

Problem Definition Logic (CodeBook)

Each analytical object has defined relationships and propagation rules.
“Down” conditions are evaluated through correlated symptoms (e.g., circuits reported as unavailable).
Competing “down problems” are compared by their explanatory strength — the one that best explains the observed symptoms is nominated as the root cause.

Examples and Correlation Scenarios

Multiple client circuits report service unavailability.
Faults in several cables (Cable 1, Cable 2, Cable 3) are candidates.
The Smarts correlation model determines which cable’s fault best explains all affected circuits, designating it as the winning root cause.
Complex test cases show that Smarts successfully correlates scattered service alarms into one accurate cable-level root cause.

Outcome

Smarts’ CodeBook-based correlation allows automatic fault localization in complex last-mile environments without direct cable monitoring.
It improves accuracy and efficiency of identifying faulty cable segments in DSL and PON networks — a task unachievable by vendor tools alone.