How Network Topology Impacts Rf Performance: A Study Powered By Graph Representation Of The Access Network (2021)

By Maher Harb, Karthik Subramanya, Ramya Narayanaswamy, Sanket Walavalkar & Dan Rice, Comcast

We have recently embarked on a project with the aim of capturing all the building blocks of the access network, their relationships, and their properties in a graph representation encompassing vertices and edges. This representation is to be available in a high performance and scalable graph database that allows access to the data through application programming interface (API) endpoints and in batch. The graph database mirrors the dynamic nature of the network by getting updated as customers get connected &disconnected, optical nodes get segmented, network equipment gets commissioned & decommissioned, as well as the happening of any other impactful network change. Having all the relational information in one source, and combining the physical & logical elements in a single view allows analyzing the access network at any level of network topology (e.g., service group, fiber node, amplifier, tap, drop) on a use case basis. The graph database technology also allows enrichment of the data with ease by overlaying device telemetry, Cable Modem Termination System (CMTS) telemetry, and maintenance data on top in order to implement algorithms for business intelligence and troubleshooting (e.g., root cause analysis).

Building the graph database required combining and reconciling data across many different sources of the organization without identified primary keys (ids) and creating algorithms to automate inference of connections. In this paper, we share Comcast’s journey into this process that is currently scaled to cover ~20% of our footprint. We present the very first use case of utilizing the network graph to study the effects of the amplifier cascade length on radio frequency (RF) performance in the upstream (US) and downstream (DS). The interest in this investigation falls within a broader question on the operational effort required to maintain nodes with large cascade of amplifiers (both in terms of depth and breadth).

Our findings reveal that, predictably, longer amplifier cascade lengths exhibit degraded RF signal-to noise ratio (SNR) -- yet with no significant impact on quality of service, likely due to the mitigating impact of the profile management application (PMA) system currently deployed in Comcast.

By clicking the "Download Paper" button, you are agreeing to our terms and conditions.

Similar Papers

Accounting for Every MHz of Bandwidth: Data & Algorithms for Artifact Discovery and Close-Packing of QAMs in Support of Spectrum Activation
By Maher Harb, Comcast; Wenlong Shen, Comcast; Matt Stehman, Comcast; Sanket Walavalkar, Comcast; Dan Rice, Comcast
A Deep Learning Approach for Detecting RF Spectrum Impairments and Conducting Root Cause Analysis
By Kevin Dugan, Justin Evans, Maher Harb; Comcast
When Physical Layer Simulation Gets Real
By Ramya Narayanaswamy, Karthik Subramanya, Dr. Richard Prodan & Larry Wolcott, Comcast
Convolutional Neural Networks for Proactive Network Management
By Jude Ferreira, Maher Harb, Karthik Subramanya, Bryan Santangelo & Dan Rice, Comcast
Graph Algorithms and Real-Time Telemetry for Intelligent Plant Operations
By Bob Lutz, Comcast; Matthew Stehman, Comcast
Detection and Classification of OFDMA Spectrum Impairments by Machine Learning
By Jude Ferreira, Comcast; Kevin Dugan, Comcast; Maher Harb, Comcast; Mike O’Dell, Comcast; Larry Wolcott, Comcast
Deploying PMA-Enabled OFDMA in Mid-Split and High-Split
By Maher Harb, Dan Rice, Kevin Dugan, Jude Ferreira, Robert Lund; Comcast
What Gets Measured Gets Done / What Gets Analyzed Gets Transformed: Analytics for a Wider/Deeper Network View
By Venk Mutalik, Dan Rice, Karthik Subramanya & Jon-en Wang, Comcast
Exploring Programmatically Generated HFC Plant Topology
By Lin Cheng, CableLabs; L. Alberto Campos, CableLabs; Justin Riggert, Comcast Corporation; Larry Wolcott, Comcast Corporation
A Machine Learning Pipeline for D3.1 Profile Management
By Maher Harb, Jude Ferreira, Dan Rice, Bryan Santangelo & Rick Spanbauer, Comcast
More Results >>