Systems and methods to differentiate access ports and uplinks in telecommunication networks using machine learning - Abstract and detailed description

Aug 4, 2026

Upload by Finn

Supervised machine learning is used to automatically distinguish uplinks from access ports in enterprise telecommunication networks. After classification, the system suppresses or reduces nonactionable noise alarms—especially critical alarms—originating from access interfaces. Network Operations Centers can apply ML-based determinations and take the required operational actions using feedback from NOC engineers. The approach reduces system noise and accelerates network operations by preventing overload of alarm-processing workflows and supports end-to-end automation via an MLOps pipeline.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Abstract

Problem statement

Solution

The End-to-end MLOps System

Network Datastore

Data Pre-processing

ML Classification (Inference)

ML Database

Role of User interface (UI)

Technical Disclosure Commons Defensive Publications Series July 2023 Systems and methods to differentiate access ports and uplinks in telecommunication networks using machine learning Anonymous Follow this and additional works at: [https://www.tdcommons.org/dpubs_series](https://www.tdcommons.org/dpubs_series) Recommended Citation Anonymous, "Systems and methods to differentiate access ports and uplinks in telecommunication networks using machine learning", Technical Disclosure Commons,(July 17, 2023) [https://www.tdcommons.org/dpubs_series/6062](https://www.tdcommons.org/dpubs_series/6062) This work is licensed under a Creative Commons Attribution 4.0 License. This Article is brought to you for free and open access by Technical Disclosure Commons. It has been accepted for inclusion in Defensive Publications Series by an authorized administrator of Technical Disclosure Commons. Systems and methods to differentiate access ports and uplinks in telecommunication networks using machine learning ABSTRACT This invention is adapted to identify uplinks vs access ports automatically using supervised Machine learning (ML) techniques. Once uplinks vs access ports are identified, the present methods automatically suppress/reduce the noise/nonactionable/non-important alarms from the access ports. Network Operations Centers (NOCs) will be able to change the uplinks vs access ports predicated by ML , and based on feedback from NOCs , required action will be taken against the access port alarms. This makes noise reduction and network operations faster and easier for the operators of enterprise networks. DETAILED DESCRIPTION Problem statement: There are tens of thousands of interfaces in a typically enterprise network. Each interface produces a variety of alarms at high rates, including link down, loss of communication, power module failure, etc. To determine whether an alarm is important or not, network engineers must take actions like clearing or acknowledging it, raising a ticket, etc. This takes time and requires a lot of knowledge from many people. Out of all interfaces (Access/Downlink + Uplink) alarms, it has been determined that 80% of them originate from access interfaces and have a severity of "Critical". Access alarms with this severity are essentially false alarms and system noise. Determining the access interfaces and suppressing the associated alarms with critical severity are the objectives of this solution. To aid in this, rule-based systems can be created, but they are difficult to maintain and incapable of handling complex situations. Hence, multi-vendor network products make every effort to automate the management of alarms. Given the very large number of interfaces, requiring network operators to manually maintain the list of access ports and uplinks is not practical. Solution: As the objective is to separate the network's interfaces/port types into access and uplink, followed by the suppression of noisy critical alarms generated by access interfaces, it is suggested using ML algorithms to do so. As a result, this problem is now a "Classification"problem, where the goal of machine learning is to categorise ports and interfaces into access and uplink ports. Once an interface is identified as an access port, an end-to-end system or pipeline will suppress the critical network alarms against it. Both, Supervised and Unsupervised ML approaches used to try to solve this problem statement and produce a reasonable result. Thus, by putting this project into action, system noise is reduced before NOC engineers are overloaded. Details of some encouraging PoC experiments can be found in the sections below. Page 1 of 13 Published by Technical Disclosure Commons, 2023 2 The End-to-end MLOps System: Several novel moving components of the end-to-end system are shown in the flow diagram (figure 1) and are also listed below. Figure 1 • Network Datastore: Data is the foundation of

FAQs

What problem does the invention address in enterprise networks?

Enterprise networks generate alarms at high rates across many interfaces. The goal is to identify which alarms are important by separating access-port alarms from uplink-related events, especially when access alarms with “Critical” severity are often false alarms and noise.

How does the system differentiate access ports from uplinks?

It formulates the task as a machine-learning classification problem and uses supervised ML techniques. The classification output determines whether an interface is an access port or an uplink.

What happens after an interface is identified as an access port?

An end-to-end system suppresses or reduces critical network alarms associated with that access port, limiting nonactionable noise and helping operators focus on actionable issues.

Aug 4, 2026

Upload by Finn