International Business Machines Corporation

Yossi Richter

Elad Yom-Tov

Yehuda Naveh

Liane Lewin-Eytan

Gidon Gershinsky

Konstantin Shagin

Daniel Patrick Connors

Donna Leigh Gresh

An advanced metering infrastructure comprises intermediate nodes. The intermediate nodes receive data from child nodes and relay a subset of the data that is not according to an expected value. The expected value may be determined based on a forecasting function computed based on past data. The expected value may be a spatial shape in an n-dimension space. A data not within the spatial shape may be considered not in accordance with the expected value. In some case, the spatial shape is defined by a centroid and a radius. The spatial shape may shift over time based on a consumption profile, such as low consumption at noon, and high consumption at evening. The consumption profiles may be determined in a learning phase, as well as shifting of spatial shapes of each group over time.

Data forecasting in a multi-node system

An advanced metering infrastructure comprises intermediate nodes. The intermediate nodes receive data from child nodes and aggregate the data according to groups of child nodes. The aggregation provides for a reduced version of the data. The reduction is performed based on groups determined by clustering. The reduced version of the data comprises data describing a group of measurements over time, such as a centroid in an n-dimension space, number of customers in the group, radius of the group and the like. The centroid may shift over time based on a consumption profile, such as low consumption at noon, and high consumption at evening. The consumption profiles may be determined in a learning phase, as well as shifting of centroids of each group over time.

Data reduction in a multi-node system

Determining a network transmitter that is more likely to cause handoff failures in a telecommunication service based on historical data records. The historical data records may be standard Call Data Records. A probability that a service provided by a first network transmitter will be handoffed to a second network transmitter is determined. An indication that a target network transmitter is overly busy is determined based on the number of failed services for each network transmitter that may handoff a service the target network transmitter and the probability that a service will be handoffed to the target network transmitter. Based on the indication, measures may be taken to increase quality level of the target network transmitter.

Detection of failures in a telecommunication system

A method for resource management includes associating respective variables with resource consumers, and identifying resources as values applicable to the variables. A group of the variables are identified as preferred variables. An assignment of the values to the variables that satisfies constraints applying to the allocation of the resources is found by repeatedly performing the steps of choosing a variable from the group, instantiating the chosen variable with a value, removing the chosen variable from the group, and pruning the domains of the other variables by propagation of the constraints. The resources are assigned to the resource consumers responsively to the assignment of the values to the variables.

Resource management using constraint programming with efficient ordering of variables

Workforce management using constraint programming with efficient ordering of variables

A computer-implemented method for resource management includes storing in a memory a set of resource consumers, resources for allocation to the resource consumers, and initial binary constraints applying to the allocation of the resources. Respective variables are associated with the resource consumers, and the resources are identified as values applicable to the variables. A non-binary scoring criterion is applied to generate scores on a graded scale for proposed assignments of one or more of the resources to respective ones of the resource consumers. One or more additional binary constraints are defined based on the graded scale of the scores. An assignment is found of the values to the variables that satisfies the initial and additional binary constraints using a constraint satisfaction problem (CSP) solver running on a processor that is coupled to the memory. The resources are assigned to the resource consumers responsively to the assignment of the values to the variables.

Growing community of inventors

Kfar Saba, Israel

Yossi Richter