Optimal Communications Systems and Network Design for Cargo Monitoring

In 2006 the Federal Bureau of Investigation (FBI) estimated that cargo theft cost the US economy between $15 and $30 billion per year. Others have noted that the indirect costs-from investigation and insurance payments-of cargo theft can be two to five times the direct losses from cargo theft. At the same time that the shipping industry is grappling with high cargo theft, exports from Asia to the USA have increased significantly resulting in bottlenecks at certain key ports on the USA's Pacific Coast. Some shipping organizations have sought to get around the bottlenecks at West Coast ports by using inland ports. To this end, they seek to offload cargo from ships directly onto trains destined for an inland intermodal traffic terminal. Once at the terminal, the freight can then be processed by Customs and then distributed within the United States. For such an effort to succeed shippers must have "visibility" into rail shipments. This dissertation studies the system trade-offs that arise when providing visibility into cargo shipments in motion. This visibility is provided through the optimal placement of sensor and communication technology. This dissertation shows that a transportation security sensor network for monitoring cargo in motion can provide timely event notification to shippers. Two generalized models--one for use when all network elements are on the train and the other for use when some are located trackside--suitable for analyzing a cargo monitoring system are presented. The models show that, under reasonable assumptions, sensor deployment reduces the overall cost of a cargo monitoring system. The models developed here enable system trade-off studies that show that the system deployment cost is inversely related to the deadline for decision maker notification. Furthermore, the system trade-off studies show that the system deployment cost is inversely related to the average train speed. The generalized models developed in this research are Mixed Integer Nonlinear Programs (MINLP). Prior research has shown that MINLP are nondeterministic polynomial time (NP) hard problems. As a result the system trade-off studies are conducted on relatively small trains with 15 units and 33 containers. Thus, a heuristic has been developed to choose the best (or close to best) way to deploy sensors to trains of arbitrary size. The heuristic has been successfully applied to a train with 105 units and 225 containers. This dissertation demonstrates that sensors and communications systems can be used to monitor cargo in motion. In addition the dissertation provides potential designers of cargo monitoring systems tools which can be used to study the trade-offs inherent in such a system. Finally, the dissertation presents a heuristic that can be used to deploy sensors to relatively large trains.

Grantor dc:publisher

University of Kansas

Year dc:date.issued

2010

Author dc:creator

• Fokum, Daniel Tangyi

Advisor dc:contributor.advisor

• Frost, Victor S.

• Computer Science
• Cargo
• Monitoring
• Sensors
• Trains
• Transportation Security Sensor Network

Statement dc:rights

• Copyright held by the author.

Language dc:language.iso

en

Dc Identifier Other

http://dissertations.umi.com/ku:11229

OAI identifier oai:identifier

oai:kuscholarworks.ku.edu:1808/37013