By Matthew Stehman, Comcast; Chris D’Andrea, Comcast; Ilana Weinstein, Comcast
As cable providers continue to develop and deploy 10G technologies, customer experience expectations are also growing in stride. These more reliable networks and faster speeds mean that even the smallest outages have a high impact on daily life. In the realm of outside plant power, the power supplies are responsible for powering the plant as well as providing back up battery power and keeping the cable plant online during commercial power outages (CPOs). Artificial intelligence (AI) can be integrated with real-time telemetry to plan resource deployments based on the expected performance of the batteries combined with predictions on customer impacts during the outage. While CPOs are likely unavoidable for multiple systems operators (MSOs), advanced techniques can be employed to mitigate potential connectivity outages during such events. The authors will discuss Comcast’s approach to minimizing customer impact during CPOs using Artificial Intelligence and Machine Learning (ML). Big Data tools and pipelines are employed to gather power supply telemetry and customer connectivity data for historical CPOs. Predictive models are built to learn from these historical events and can provide insights during active CPOs. These predictive models can then be integrated into field operations platforms to help plan work and even proactively message customers. This paper will illustrate the power of using historical power outage data with AI/ML frameworks to support the connectivity expectations of a 10G network.