Arguably the biggest challenge for the HFC plant moving forward is implementing next generation technologies in nodes that operate in a limited power consumption environment. This, for example, is the backdrop for the evolution to distributed access architecture (DAA,) full duplex (FDX), and networking nodes which are expected to require more average, and larger ranges of power consumption. There is also a discrepancy between the higher power/thermal dissipation capability of new nodes and the maximal power that is allotted to them per multiple system operator (MSO) product specifications. This creates an opportunity for an analytics-based application of power management that maximizes the performance capabilities of nodes but at the same time keeps in check or reduces the consumption of their system power footprint.
Because of MSO desires to avoid adding new power supplies when deploying additional nodes as part of, for example, fiber-deep deployment, nodes are now part of a power consumption cluster, which is effectively a collection of nodes serviced by a power supply where ultimately the power envelope that matters is that for the collection of nodes in the cluster, and not of any single node in particular. In effect, nodes are reverting back to a more centralized powering schemes from their previous distributed powering architectures, where the power supply placement is often no longer optimal due to new nodes being added downstream of the original node. Significant additional losses in power due to the Joule heating or I2R losses in the coax used to transport power to the new node locations are now added to the powering requirements of the new nodes themselves. The new power consumption means that many power supplies may become challenged to supply sufficient power as new devices such as wireless strand-mounted devices are added to the HFC plant.
Therefore, power sensory information of node function is now necessary. Sensors that are hardware parts can be added to nodes to make the reading and reporting of power state information possible. This sensory information can be collected and maintained centrally, within a general cloud infrastructure.
Making such energy consumption information available from sensors allows for an analytic comparison and optimization of power and/or performance settings of elements in the cluster to optimize performance while improving energy efficiency. And finally, in keeping with the current trend for increasing intelligence in network operations, these sensors and associated data can be the inputs to machine learning algorithms that provide predictions and necessary decisions to optimize or evolve a system and thereby facilitate introduction of new or different elements into the cluster.
This paper is a novel proposition for an analytics-based application that manages the problem of wide deployment of new technology in nodes. In this paper we describe the hardware, sensory capability expected, the cloud-based architecture needed for data collection, storage and analysis, the logic applied to analytical engines, and the process for execution of optimal states in the presence of a broader policy mechanism, and finally the inclusion of a machine learning principles for integration of new technologies as nodes evolve.