검색 상세

Mobility Control of LTE-R for Enhanced Safety of Automated High-speed Railway Control

Mobility Control of LTE-R for Enhanced Safety of Automated High-speed Railway Control

초록/요약

This thesis targets enhanced reliability of LTE-R wireless networks for automatic HSR control. In a typical railway control system, numerous control entities are cooperating to maintain safe and efficient railway transportation service by message exchanges. Control servers understands overall railway status by measurement data sent from sensor nodes dispersed along railways. On the other hand, the control servers send command packets to actuators dispersed along the railways as well. The actuators yield physical effects. To provide connectivity for sensor and actuators dispersed in broad area, LTE-R functions as a wayside access network, and LTE-R is again connected to a wired network where the control servers reside in. As a result, the wayside devices can communicate with the control servers only through LTE-R. Therefore, safe railway transportation service depends on reliable operation of LTE-R’s data delivery service. Current LTE-R inherits most of its technical features from general LTE including mobility control. Because of its advantages in high reliability, efficiency in bandwidth utilization, stable connectivity, general LTE is widely accepted as wireless access network for voice, video, and data services. However, merely inheriting technical features from LTE to LTE-R is problematic since railway environment is different from the environment assumed in the general LTE. This thesis analyzes problems of LTE-R’s mobility control and tries to resolve the problems with machine learning algorithms. The first problem is vulnerability of the LTE-R to signaling attack. LTE-R signaling attack seeks to consume abundant amount of resource of control plane of LTE-R exploiting vulnerability of mobility control mechanism. Signaling attack leads to degenerate quality of LTE-R’s data communication. As a result, real-time railway control becomes unavailable. This thesis proposes an LTE-R signaling attack detection scheme based on traffic modeling by Hidden semi-Markov Model which is a machine learning algorithm. It is verified by simulation results that the proposed scheme outperforms a current scheme with more accurate detection. The second problem is inadequate handover decision algorithm by LTE-R. For a train, a mobile relay relays data communication between a wayside access point, called DeNB, and an onboard steering device, train controller. Handover decision algorithm in general LTE holds handover initiation for a mobile station until the mobile station has resided in non-serving DeNB. Since mobile relays on a train moves with high-speed, the standard handover decision algorithm would make mobile relay retain week wireless connection. Thus, the mobile relay’s should relay data packets with week wireless signal when the mobile relay is moving around cell boundaries, and the reliable communication for train controllers become unable to guaranteed. This thesis proposes a handover decision algorithm for LTE-R based on a machine learning algorithm, Bayesian regression. With simulation results, this thesis verifies that the proposed scheme achieves stronger signal strength for mobile relays and enhanced packet delivery ratio as well.

more

목차

Chapter I Introduction 1
Chapter II Background 4
II.A Network based automatic HSR control system 4
II.B Requirement of LTE-R 8
II.C LTE-R mobility control 9
II.C.1 Inactive mobile station 11
II.C.2 Active mobile station 13
II.D Problems of LTE-R mobility control 16
II.D.1 Risk of degraded railway control by LTE signaling attack 17
II.D.2 Interruption of railway control by inappropriate LTE-R handover decision 18
Chapter III Machine-learning based Anomaly Detection for Alleviating Risk of Degraded Railway Control by LTE-R Signaling Attack 21
III.A LTE signaling attack 22
III.B Related work 23
III.C Overview 26
III.D Mathematical model for wakeup packet generation 31
III.E Proposed signaling attack detection algorithm 37
III.F Simulation Results 40
III.F.1 Simulation setup 40
III.F.2 Results 42
III.G Chapter summary 53
Chapter IV Machine Learning Based Handover Initiation for Seamless Train Control 54
IV.A Inappropriate handover decision of LTE-R 55
IV.B Related work 57
IV.C Overview 59
IV.D Mathematical model for cell boundary crossing time prediction 62
IV.D.1 Why Bayesian Regression Model 63
IV.D.2 Bayesian regression model selection 67
IV.D.3 Gaussian regression based RCDT prediction 73
IV.E Proposed handover decision algorithm 75
IV.E.1 Algorithm description 75
IV.E.2 Analysis on computation complexity 81
IV.F Simulation Results 83
IV.F.1 Simulation Setup 83
IV.F.2 Results 85
IV.G Chapter summary 95
Chapter V Conclusions 97
Bibliography 100

more
반출 Meta View 목록