Abstract: Devices, computer-readable media and methods are disclosed for selecting paths in reconfigurable optical add/drop multiplexer (ROADM) networks using machine learning. In one example, a method includes defining a feature set for a proposed path through a wavelength division multiplexing network, wherein the proposed path traverses at least one link in the network, and wherein the at least one link connects a pair of reconfigurable optical add/drop multiplexers, predicting an optical performance of the proposed path, wherein the predicting employs a machine learning model that takes the feature set as an input and outputs a metric that quantifies predicted optical performance, and determining whether to deploy a new wavelength on the proposed path based on the predicted optical performance of the proposed path.

Abstract: Devices, computer-readable media and methods are disclosed for selecting paths in reconfigurable optical add/drop multiplexer (ROADM) networks using machine learning. In one example, a method includes defining a feature set for a proposed path through a wavelength division multiplexing network, wherein the proposed path traverses at least one link in the network, and wherein the at least one link connects a pair of reconfigurable optical add/drop multiplexers, predicting an optical performance of the proposed path, wherein the predicting employs a machine learning model that takes the feature set as an input and outputs a metric that quantifies predicted optical performance, and determining whether to deploy a new wavelength on the proposed path based on the predicted optical performance of the proposed path.

Abstract: Devices, computer-readable media and methods are disclosed for selecting paths in reconfigurable optical add/drop multiplexer (ROADM) networks using machine learning. In one example, a method includes defining a feature set for a proposed path through a wavelength division multiplexing network, wherein the proposed path traverses at least one link in the network, and wherein the at least one link connects a pair of reconfigurable optical add/drop multiplexers, predicting an optical performance of the proposed path, wherein the predicting employs a machine learning model that takes the feature set as an input and outputs a metric that quantifies predicted optical performance, and determining whether to deploy a new wavelength on the proposed path based on the predicted optical performance of the proposed path.

Abstract: Devices, computer-readable media and methods are disclosed for selecting paths in reconfigurable optical add/drop multiplexer (ROADM) networks using machine learning. In one example, a method includes defining a feature set for a proposed path through a wavelength division multiplexing network, wherein the proposed path traverses at least one link in the network, and wherein the at least one link connects a pair of reconfigurable optical add/drop multiplexers, predicting an optical performance of the proposed path, wherein the predicting employs a machine learning model that takes the feature set as an input and outputs a metric that quantifies predicted optical performance, and determining whether to deploy a new wavelength on the proposed path based on the predicted optical performance of the proposed path.

Abstract: Devices, computer-readable media and methods are disclosed for selecting paths in reconfigurable optical add/drop multiplexer (ROADM) networks using machine learning. In one example, a method includes defining a feature set for a proposed path through a wavelength division multiplexing network, wherein the proposed path traverses at least one link in the network, and wherein the at least one link connects a pair of reconfigurable optical add/drop multiplexers, predicting an optical performance of the proposed path, wherein the predicting employs a machine learning model that takes the feature set as an input and outputs a metric that quantifies predicted optical performance, and determining whether to deploy a new wavelength on the proposed path based on the predicted optical performance of the proposed path.

Abstract: Devices, computer-readable media and methods are disclosed for selecting paths in reconfigurable optical add/drop multiplexer (ROADM) networks using machine learning. In one example, a method includes defining a feature set for a proposed path through a wavelength division multiplexing network, wherein the proposed path traverses at least one link in the network, and wherein the at least one link connects a pair of reconfigurable optical add/drop multiplexers, predicting an optical performance of the proposed path, wherein the predicting employs a machine learning model that takes the feature set as an input and outputs a metric that quantifies predicted optical performance, and determining whether to deploy a new wavelength on the proposed path based on the predicted optical performance of the proposed path.

Abstract: In a tool for use by network planners in designing metropolitan IP aggregation networks, several optimization methods are integrated into a tool. A k-means algorithm is used to choose access packet switch locations. A dual-k-means algorithm is used to choose the backbone switch locations. Each access packet switch is dual homed to two backbone packet switches using two diverse paths. The diverse path configuration is found using a maxflow-mincost algorithm on a modified fiber map topology. The link topology connecting the backbone packet switches to each other is designed using a heuristic that creates a skeleton network topology and then adds express links one by one, testing each link to assure reduced overall network cost. The resulting network topology is then improved upon by local search.

Abstract: A methodology for optimizing the cost associated with access charges incurred by network service providers when leasing communications facilities from a local service provider has been developed that incorporates physical reconfiguration of the access network topology as a cost-saving measure, where and when appropriate. The methodology identifies opportunities to reduce access charges by performing the following functions: (1) moving T1 facilities from one hub location to another; (2) moving T1 facilities from one T3 facility to another (at the same hub location) in order to disconnect unneeded T3 facilities; and (3) adding new T3 facilities in order to reduce overall access expenses and/or to provide new capacity for future growth, including adding potential new hub locations. The methodology is appropriate for use with various types of network facilities, T1 and T3 being considered as exemplary only.

Abstract: A methodology for optimizing the cost associated with access charges incurred by network service providers when leasing communications facilities from a local service provider has been developed that incorporates physical reconfiguration of the access network topology as a cost-saving measure, where and when appropriate. The methodology identifies opportunities to reduce access charges by performing the following functions: (1) moving T1 facilities from one hub location to another; (2) moving T1 facilities from one T3 facility to another (at the same hub location) in order to disconnect unneeded T3 facilities; and (3) adding new T3 facilities in order to reduce overall access expenses and/or to provide new capacity for future growth, including adding potential new hub locations. The methodology is appropriate for use with various types of network facilities, T1 and T3 being considered as exemplary only.

Abstract: A methodology for optimizing the access charges paid by a long distance carrier involves the use of a “sizing” operation to determine the optimal balance between direct trunks and access tandems (on a per LATA basis) so as to provide minimal access charges. Once the sizing has been completed, a mixed integer program is used to find the optimal MUX assignments for each LATA. The sizing operation is then re-visited, based on the MUX assignments, to determine if adjustments are required to reduce costs. Any changes in the sizing are then reviewed in the MUX assignment mixed integer program. The end result is a definition of the quantity of direct trunks, access tandem trunks and overflow minutes-of-use charges for each end office, as well as the number of high capacity facilities required for each MUX.

Abstract: A method and system are provided for computing an optimal restoration capacity and/or optimal restoration paths for a network to resolve a restoration scenario by solving a linear program (LP) model. The LP model includes decision variables corresponding to restoration capacity and restoration paths and constraints requiring restoration of traffic and conservation of capacity in the network. For example, the system determines a set of network paths that need to be restored in the network, preprocesses network data corresponding to the network to reduce LP processing workload and time; generates possible restoration paths and solves the LP model, preferably through column generation methods, to determine the optimal restoration capacity and/or the optimal restoration paths to resolve the restoration scenario; changes the LP solution into integer form, as necessary, and changes the LP solution in integer form to the original format of the network data or an equivalent thereof.