Autonomic DOCSIS Networking

Schnitzer, Jason Kazimir

Graduate Thesis Or Dissertation

Autonomic DOCSIS Networking Public Deposited

Analytics

Citations

Citeable URL: https://scholar.colorado.edu/concern/graduate_thesis_or_dissertations/gq67js532

Abstract

Now in its sixth generation, the DOCSIS protocol continues to play a significant role in the global Internet infrastructure. However, today, the scale and complexity of these broadband access networks have surpassed the human capacity to oversee them. Conventional DOCSIS network management and operations rely on deeply embedded assumptions rooted in legacy protocols, static information models, and indurated system architectures - forcing limitations in both visibility and control. These deficiencies conspire to reduce network availability, decrease performance, increase operational costs, and ultimately degrade the overall quality of the connected experience.

This thesis addresses these deficits by presenting a principled approach to the self-regulation of DOCSIS networks. We propose Autonomic DOCSIS Networking, which enables self-management of specific network functions through the application of emerging programmable protocols, analysis methods, and system architectures. This new approach reduces dependency on human supervision by introducing closed-loop network control - an adaptive system of collection, analysis, and actuation purposed to achieve desired operational outcomes in complex access network technology deployments of significant scale.

To demonstrate the model of Autonomic DOCSIS Networking, two novel studies of self-management capability are presented. The first implements self-optimization of physical layer (layer-1) OFDM channel transmission parameters to maximize network throughput and error performance. In the second, the self-configuration of media access control (layer-2) resources is shown to improve the forwarding performance of wireless backhaul traffic over DOCSIS segments in the evidence of network congestion.

Creator