Federated Learning Based Proactive Handover in Millimeter-wave Vehicular Networks

TL;DR

This paper introduces a federated learning framework for proactive handover in millimeter-wave vehicular networks, aiming to reduce handover delays and improve user QoS by leveraging distributed learning.

Contribution

It presents a novel federated learning approach tailored for vehicular networks that handles limited user storage, increases participant numbers, and adapts to mobility patterns.

Findings

Reduces unnecessary handovers compared to reactive schemes

Improves quality of service for mobile users

Validates effectiveness through simulations

Abstract

Proactive handover can avoid frequent handovers and reduce handover delay, which plays an important role in maintaining the quality of service (QoS) for mobile users in millimeter-wave vehicular networks. To reduce the communication cost of training the learning model for proactive handover, we propose a federated learning (FL) framework. The proposed FL framework can accommodate the limited storage capacity of each user, increase the number of users who participate in the FL, and adapt to the dynamic mobility pattern. Simulation results validate the effectiveness of the proposed FL framework. Compared to reactive handover schemes, the proposed handover scheme can reduce unnecessary handovers and improve the QoS of users simultaneously.