At&t Intellectual Property I, LP

Velin Kounev

Arun Jotshi

Yaron Kanza

Krystian Czapiga

Philip E Brown

Gaurav Thakur

Weiwei Duan

Poornima Suresh

Kathleen S Meier-Hellstern

Gopalakrishnan Meempat

Danielle Liu

James Klosowski

Philip M Brown

Fei Wu

Shang Li

Serkan Isci

Divith Aruni Babu

In one example, a method performed by a processing system including at least one processor includes creating a geospatial model of an environment in which a cellular network is to be deployed, transforming, for each cellular antenna of a proposed antenna layout of the cellular network, a radiation pattern of the each cellular antenna into a signal strength array, to create a plurality of signal strength arrays, augmenting, for each signal strength array of the plurality of signal strength arrays, the each signal strength array with at least one parameter of a corresponding cellular antenna of the proposed antenna layout and at least one value describing the environment in which the cellular network is to be deployed, and estimating a coverage of the proposed antenna layout based on the signal strength array, as augmented, using a machine learning model.

Machine-learning-based wireless planning using antenna radiation patterns

An example method performed by a processing system includes obtaining a light detecting and ranging point cloud of a building, where the point cloud includes a plurality of points, and where each point is associated with a set of (x,y,z) coordinates. A first point of the plurality of points is assigned to a subset of the plurality of points that is associated with the building, where the subset includes points whose (x,y) coordinates fall within a footprint of the building. The first point is grouped into a first cluster according to at least one of: a (z) coordinate of the first point and a gradient to which the first point belongs. A first prism formed by the first cluster is constructed. A model of the building is stored as a plurality of connected prisms, where the plurality of connected prisms includes the first prism.

Constructing compact three-dimensional building models

Aspects of the subject disclosure may include, for example, a process for selecting equipment locations such as of cellular antennas, based on a combination of a geospatial grid representation of a planning area and optimization algorithms (which can be combined with propagation models and a 3D model of the world) where the optimization algorithm can select a deployment from a large space of options and would make RAN planning much more efficient. Other embodiments are disclosed.

RAN planning using grid-based optimization

A processing system may obtain usage volume information for endpoint devices for at least one cell site of a cellular network, determine at least one earning value of the at least one cell site based upon a summation of an earning metric of each of the endpoint devices for the at least one cell site, the earning metric comprising for each of the endpoint devices in a given time period: a total earning for the cellular network from the endpoint device times a ratio of the usage volume via the at least one cell site divided by the total usage volume via the cellular network, train a prediction model to predict an earning value of a new cell site, based upon geospatial features of the at least one cell site as predictor factors, and determine a predicted earning value of the new cell site via the prediction model.

Geospatial-based forecasting for access point deployments

Aspects of the subject disclosure may include, for example, network deployment or radio-propagation computation based on a combination of photon mapping and machine learning including supporting near-real-time computation of the radio transmissions for different layouts of antennas and allowing examination of a large variety of antenna locations and layouts, changing configuration details, e.g., tilting antennas or optimally selecting the sector that each antenna covers, and so on. Other embodiments are disclosed.

Real-time ML-supported radio propagation computation for RAN planning

Aspects of the subject disclosure may include, for example, obtaining user input identifying a first user-identified network feature of a training image of a geographical region. The training image and the user-identified feature are provided to a neural network adapted to train itself according to the user-identified features to obtain a first trained result that classifies objects within the image according to the user-identified feature. The training image and the first trained result are displayed, and user-initiated feedback is obtained to determine whether a training requirement has been satisfied. If not satisfied, the user-initiated feedback is provided to the neural network, which retrains itself according to the feedback to obtain a second trained result that identifies an updated machine-recognized feature of the training image. The process is repeated until a training requirement has been satisfied, after which a map is annotated according to the machine-recognized feature. Other embodiments are disclosed.

AI-based, semi-supervised interactive map enrichment for radio access network planning

Architectures and techniques are presented that improve or enhance a network planning procedure such as by selecting a more efficient test buffer that is used to identify objects that might intersect a Fresnel zone between two transceivers. An improved test buffer (e.g., buffer region) can be one that is constructed from a plurality of rectangles situated along a line of sight of the two transceivers and that are oriented according to cardinal directions.

Adaptive buffer region for line-of-sight network planning

Aspects of the subject disclosure may include, for example, calculating a Fresnel zone about a line of sight (LoS) between a pair of antennas positioned on antenna mounts. The Fresnel zone is projected onto a geospatial grid of 3D cells, each having a height above a horizontal plane. Cells that intersect the Fresnel zone projection are selected to obtain a subset cells with constraint heights determined according to heights of the subset cells that are adjusted according to the Fresnel zone. The LoS is revised according to an algorithm to obtain a set of updated LoS. Those updated LoS that do not intersect the set of adjusted constraint heights are identified as solutions that include an optimal solution, should one exist. Other embodiments are disclosed.

System and method for geospatial planning of wireless backhaul links

Aspects of the subject disclosure may include, for example, identifying a first object included in at least one image in accordance with an execution of an image processing algorithm, analyzing a plurality of parameters in accordance with at least one model responsive to the identifying of the first object included in the at least one image, wherein each parameter of the plurality of parameters is associated with the first object or a second object, selecting one of the first object or the second object for receiving at least one communication network resource responsive to the analyzing of the plurality of parameters, wherein the selecting results in a selected object, and presenting the selected object on a presentation device. Other embodiments are disclosed.

Growing community of inventors

Weehawken, NJ, United States of America

Velin Kounev