By Salvatore (Sam) Torrente, Petar Djukic, Dmitri Fedorov, Mehran Bagheri & Marco Naveda, Ciena
Network operators understand that an increasing number of discrepancies between their network and inventory databases can lead to (1) greater service fallout rates, (2) impact to customer experience and (3) potential significant loss of revenue from dissatisfied customers. While networks are evolving toward more “addressable” (e.g., discoverable) devices, there is still a significant amount of resource elements in the network that are non-addressable (e.g., patch panels, fibers) thus requiring manual tracking in databases. Furthermore, most existing inventory systems either do not, or cannot, stay in lockstep with the “as-is” state of the network and continue to impose on network operators a significant amount of error prone manual tasks. While network-to-inventory “manual audits” can provide a snap-shot update in time, each audit can cost network operators millions of dollars in consulting/staff costs and expenses they can no longer afford. And, with networks and services evolving quickly to keep up with customer demand this only increases the number of yearly audits needed to stay up to date with the network. As a result, network operators find it increasingly difficult to reconcile their physical network with their network inventory databases leading to (1) increasing number of discrepancies that is no longer manageable through traditional manual corrective means; and (2) generating an increasing amount of stranded assets that may no longer generate revenue to the network operator while still consuming space, power, and HVAC expenses.
In this paper, we will describe an architecture that uses augmented reality (AR) and artificial intelligence (AI) as part of a planned network maintenance or upgrade to reconcile the physical network with planned inventory. We propose a new inventory architecture in which AR is used to capture images of network devices, seamlessly and unobtrusively, in the field during regular maintenance tasks, which are then analyzed using AI to update the inventory. It will also describe how technicians are assisted in verifying that a task was performed correctly. Visual examples will show, in tutorial style, how the technologies blend together.