By Venk Mutalik, Dan Rice, Karthik Subramanya & Jon-en Wang, Comcast
With the acceleration of technology in homes comes a corresponding increase in the number of switching power supplies potentially impacting the upstream plant. More and more in-home electronics devices -Internet-connected appliances, battery chargers, LED lights, video set-top boxes (STBs), broadband gateways and cable modems -- come with switching power supplies inside of them, which contribute to an age-old issue known as Common Mode Disturbance, or CMD.
This is happening coincident with the industrial shift away from traditional centralized architectures to distributed architectures. Distributed Access Architectures (DAA) are on the rise because of a growing need to fulfill newer needs, such as low-latency and high-speed applications. Yet the traditionally persistent issue that is Common Mode Disturbance (CMD) continues to impact networks in negative ways. While tools have improved dramatically in the last few years in addressing such pesky problems as CMD, it continues nonetheless to impact even the more modern fiber deeper and distributed networks.
Specifically, the rise of CMD noise, in part trigged by the explosion of Internet-connected CPE in our customers’ homes, catalyzed within Comcast an impairment identification and mitigation framework described in this paper. The “identification” portion of the framework is informed by machine-level telemetry data, to better measure the impairment; and the mitigation portion of the framework is enabled by advanced data analysis. (Hence the title, “what gets measured gets done; what gets analyzed gets transformed.”) Our intent is to provide new insights into age-old problems, as well as a framework for analyzing old and new problems alike. New, machine-informed ways of looking at the traditional time and frequency domains of RF information can help to create a “Taxonomy of RF Impairments” – a first in the industry (to our knowledge), which has developed as a playbook to tackle impairments. This work will ultimately lead the industry toward an effective use of cable assets and aid in the creation of a more elastic, low latency network