For nearly as long as DOCSIS has been deployed in cable networks, capacity planning has been a challenge for operators. As the nature of traffic has evolved across the IP network, the accuracy and usefulness of traditional DOCSIS capacity modelling techniques has diminished.
This paper describes an alternative, simpler model for predicting the peak capacity of a DOCSIS network, as measured bidirectionally through the CCAP (converged cable access platform) core, solely based on the number of subscribers attached to the DOCSIS platform and the demographics of the customers. This model is based on analysis of current traffic patterns and traffic types and subscriber behaviors within DOCSIS networks, which is presented herein. Core peak data was sampled over multiple months and across dozens of systems, and is used to show that the accuracy of the technique is superior to other modelling methods.