Methods and apparatus for using machine learning to facilitate network handoffs between access points

Abstract

Data is collected at access points about mobile communications systems, e.g., vehicles, use of access points and the quality of the connections that are provided. The collected information is provided to a machine learning access point (MLAP) model training device that generates one or more MLAP prediction models with each model corresponding to a geographic area in which access points are located. Vehicles provide location, acceleration and/or other information, such as a user entered destination. A vehicle request for an AP recommendation includes this information which allows the AP prediction engine to predict a path of travel. The MLAP prediction engine uses the stored models and/or information about AP loading, to predict what AP a vehicle requesting an AP recommendation should attach to next. The predicted AP is returned as a handoff recommendation along with the location and/or time the vehicle should initiate the handoff to the recommended AP.