Reinforcement Learning Handover for Drone Mobility

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current mobility management procedures face challenges in providing robust mobility support for three-dimensional (3D) coverage, particularly for low altitude drones, due to their arbitrary trajectories and faster movement speeds compared to ground users.

Innovation Solution

A machine-learning assisted mobility management procedure using reinforcement learning is employed to optimize handover decisions and drone trajectories, incorporating real-time RSRP/RSRQ data, UE speed, and 3D trajectory information to minimize handovers, radio link failures, and signaling overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional handover mechanisms are used for drone mobility management, then basic connectivity is maintained, but handover frequency increases and signaling overhead increases due to arbitrary trajectories and faster movement speeds

Engineering Contradiction:

Improvedrone connectivityVSAvoidhandover frequency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by predicting future drone positions and pre-determining handover targets before handover is actually needed. The machine learning model forecasts drone trajectory and identifies optimal target base stations in advance, allowing the network to prepare handover parameters and resources beforehand, thus reducing actual handover execution time and frequency.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements dynamic handover parameters that adapt to drone movement characteristics. The machine learning model continuously learns from drone trajectory patterns, speed variations, and signal conditions to dynamically adjust handover thresholds, timing, and target selection criteria, enabling the system to optimize handover frequency for each specific drone flight pattern rather than using static parameters.

Inventive Principle:

Principle #15Dynamics

2Reliability

If traditional handover mechanisms are used for drone mobility management, then basic connectivity is maintained, but signaling overhead increases due to frequent handovers

Engineering Contradiction:

Improvedrone connectivityVSAvoidsignaling overhead

Core Design Contradiction:

ReliabilityVSObject-generated harmful factors

Solution Approach 1:

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system implements a feedback mechanism where the machine learning model continuously learns from actual handover outcomes, signal quality measurements, and drone movement patterns. This feedback loop allows the system to refine its predictions and adjust handover decisions to minimize unnecessary handovers and reduce signaling overhead while maintaining connectivity reliability.

Inventive Principle:

Principle #23Feedback

3Productivity

If reinforcement learning is used to optimize handover decisions, then handover efficiency improves and ping-pong handovers are reduced, but system complexity increases

Engineering Contradiction:

Improvehandover efficiencyVSAvoidsystem complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The system introduces a machine learning model as an intermediary layer between raw measurement data and handover decision-making. This intermediary processes complex trajectory predictions, signal quality assessments, and handover optimizations, shielding the core network from complexity while enabling sophisticated handover management. The ML model acts as a smart intermediary that translates complex inputs into optimized handover recommendations.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentEP4038960B1Efficient determination of handover operations using reinforcement learning

Publication Date: 2025.05.07 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

Figure 1
Figure 2
Figure 3

AI summary

According to some embodiments, a method performed by a network node for mobility management comprises obtaining data samples for modeling a wireless network environment that comprises a plurality of cells and building a machine learning model of the wireless network using the obtained data samples. The machine learning model is trained to determine a sequence of handovers for a wireless device among the plurality of cells for the wireless device to traverse from a source cell to a destination cell. The method further comprises receiving mobility information for a wireless device, determining one or more handover operations for the wireless device based on the mobility information, and transmitting the one or more handover operations to the wireless device.