ML-Based Ping-Pong Offset Prediction for Cellular Handovers

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current mobility robustness optimization algorithms in cellular mobile communications fail to effectively prevent ping-pong handovers, leading to unnecessary handovers between serving cells, which result in mobility-related failures and resource wastage.

Innovation Solution

A machine learning-based technique dynamically predicts and tunes the Ping-Pong Offset (PPOffset) after a handover, using user equipment trajectory, speed, and received signal levels to determine whether to execute a handover back to the previous serving cell, thereby avoiding ping-pong handovers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional mobility robustness optimization algorithms are used to control handover parameters, then handover procedures can be standardized, but ping-pong handovers cannot be effectively prevented

Engineering Contradiction:
Improvehandover stabilityVSAvoidnetwork resource efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent applies dynamics by making the PPOffset parameter adaptive rather than static. The machine learning model dynamically adjusts the PPOffset value based on real-time inputs including UE speed, trajectory, and received signal levels. This dynamic adjustment allows the system to prevent ping-pong handovers in specific scenarios while maintaining standard handover procedures in normal conditions, thus resolving the contradiction between handover stability and network resource efficiency.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter PPOffset from a fixed network-configured value to a dynamically predicted value based on machine learning. By continuously adjusting this parameter according to UE mobility characteristics and signal conditions, the system prevents unnecessary handovers without compromising the reliability of legitimate handover procedures, thereby improving both handover stability and network resource efficiency.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If machine learning based dynamic PPOffset prediction is implemented, then ping-pong handovers can be avoided, but system complexity increases

Engineering Contradiction:
Improvehandover stabilityVSAvoidprediction system complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent introduces a machine learning model as an intermediary between the UE and the network node. This intermediary predicts the optimal PPOffset value based on UE mobility characteristics and signal conditions, then provides this prediction to the network for handover decision-making. This intermediary approach enables intelligent ping-pong prevention without requiring complex changes to the core handover mechanism, thus managing system complexity while improving handover stability.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The UE performs self-service by autonomously collecting its own mobility data (speed, trajectory) and signal measurements, then feeding this information to the machine learning model for PPOffset prediction. This self-service approach eliminates the need for complex network-side monitoring and control mechanisms, reducing overall system complexity while achieving reliable ping-pong handover prevention.

Inventive Principle:
Principle #25Self-service

3Ease of operation

If handover decisions are made based on signal levels alone, then decision process is simple, but unnecessary handovers occur

Engineering Contradiction:
Improvehandover decision simplicityVSAvoidhandover accuracy
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent applies preliminary action by having the machine learning model predict the optimal PPOffset value in advance, before the actual handover decision is made. This prediction incorporates UE speed, trajectory, and signal characteristics to pre-determine the appropriate offset value. When a handover is considered, this pre-calculated PPOffset is applied to the signal level comparison, preventing unnecessary handovers while maintaining a simple decision process. This resolves the contradiction by adding predictive intelligence without complicating the actual handover execution.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4346278A1Using maching learning to avoid ping-pong handover
Publication Date: 2024.04.03 NOKIA TECHNOLOGIES OY
  • EP4346278A1 patent drawingFigure 1
  • EP4346278A1 patent drawingFigure 2
  • EP4346278A1 patent drawingFigure 3

AI summary

Example embodiments of the invention provide at least a method and apparatus to perform communicating by a network node with a user equipment of a communication network a ping-pong offset prediction request message; determining a ping-pong offset prediction wherein the ping-pong offset prediction is taking into account a change in at least one of a speed, trajectory, or received signal levels from a previous serving cell; based on the determining, sending to the network node a ping-pong offset prediction, and wherein based on the ping-pong offset prediction the handover back to the previous serving cell is one of executed or not executed.