I. INTRODUCTION
EXPERTS estimate that the railway industry will receive
US$300 billion worth of global investment for development,
upgrading, and expansion over the five years from
2009 [43]. Ollier [98] noted that effective management of rail
infrastructure will be vital to this development, upgrading, and
expansion, particularly if coupled with a move to intelligent
infrastructure [39]. A key part of the management will be condition
monitoring. Condition monitoring detects and identifies
deterioration in structures and infrastructure before the deterioration
causes a failure or prevents rail operations. In simple condition
monitoring, sensors monitor the condition of a structure
or machinery. If the sensor readings reach a predetermined limit
or fault condition, then an alarm is activated. However, this sim-plistic approach may lead to a large number of false alarms and
missed failures [36]. It only provides local analysis but does not
take advantage of the superior capabilities when the sensors are
networked and their data processed collectively. Integrated data
processing allows an overall picture of an asset’s condition to be
achieved and overall condition trends to be determined [97].