Optimizing a cellular network using machine learning

Abstract

This document describes techniques and apparatuses for optimizing a cellular network using machine learning. In particular, a network-optimization controller (170) uses machine learning to determine an optimized network-configuration parameter (460) that affects a performance metric (410) of the cellular network. To make this determination, the network-optimization controller (170) requests and analyzes gradients (440) determined by one or more user equipments (UEs), one or more base stations, or combinations thereof. By using machine learning, the network-optimization controller (170) identifies different optimized network-configuration parameters (460) associated with different local optima or global optima of an optimization function, and selects a particular optimized network-configuration parameter (460) that is appropriate for a given environment. In this manner, the network-optimization controller (170) dynamically optimizes the cellular network to account for both short-term and long-term environmental changes.

Description

Background technique [0001] Wireless network providers perform cell planning to design cellular networks. Through cell planning, the network provider determines the number of base stations to be deployed, the location of these base stations, and the configuration of these base stations to achieve a specific coverage area, quality of service or operating cost. To make these determinations, the cell planning process analyzes the geographic area to determine expected traffic based on terrain and clutter in the geographic area and to simulate signal propagation characteristics. [0002] However, the analysis and simulations used for cell planning may assume a static nominal environment (eg, a particular population density, a particular terrain, a particular clutter or land use classification, or a particular type of weather). Therefore, the design and operating configuration of the cellular network determined by the cell plan may not be optimal for environments other than the ass...

AI Technical Summary

Problems solved by technology

Additionally, where cell planning is time-intensive or costly, it may not be feasible to periodically perform cell planning to update the design or operational configuration of the cellular network based on these environmental changes

Therefore, it is challenging to dynamically optimize cellular networks and take into account both short-term and long-term environmental changes

Images