The remote upstream (US) scheduler is a low latency Remote PHY solution proposed in Section 1 to maintain low latency for DOCSIS regardless of the CIN distance. Moving the US scheduler from the CCAP core to the remote PHY device (RPD) creates a new management interface between the remote US scheduler at the RPD and the upper MAC layer clients at the CCAP core. This entails a robust and standard application programming interface (API) for easy configuration / management of multiple remote US schedulers at scale and enabling interoperability between different CCAP core and RPD venders.
Towards this goal, we propose a YANG based API to manage the remote US scheduler. YANG is an API contract language widely used in the world of networking and is now being introduced into the world of DOCSIS. A specification written in YANG is referred as a “YANG module”, and a set of YANG modules are collectively called a “YANG model”. A YANG model characterizes the behavior of a network function with data hosted by the server that a client can manipulate and observe using standardized operations. Once the YANG model is published, both client and server can have faith that the other knows the syntax and semantics behind the modeled data.
In this paper, we describe how we used YANG to define the remote US scheduler service contract. Our YANG model has several modules. The first is the US scheduler itself that provides the scheduling services such as best effort, rtPS, nrtPS, UGS and PGS. The second component is the MAP Builder that that allocates bandwidth across the US RF channels. The remote upstream YANG model is intended to be included in the CableLabs’ standard modules and become an integral part of the standard DOCSISYANG ecosystem.
The rest of the paper is organized as follows. Section 2 explains why the remote US scheduler may help maintain low-latency for DOCSIS in R-PHY deployments. Section 3 shows how to separate the real-time scheduling functions from the CCAP core while keeping DOCSIS control plane intact. Section 4 examines the reasoning for using YANG data model-driven management for the remote US scheduler.
Section 5 explains the fundamental modeling principles to establish the remote US scheduler YANG model. Section 6 presents the remote US scheduler YANG model structure and the key components.
Section 7 discusses how to add the YANG based remote US scheduler to existing RPDs that are managed by the legacy Generic Control Protocol (GCP). Section 8 describes how to integrate the remote US scheduler into the upcoming DOCSIS YANG ecosystem. Finally, Section 9 concludes the paper and highlights the takeaways.