Our goal: To analyze the effects using different mobility models in VANET simulations on privacy metrics.
We do this through simulation and analysis of several mobility models that could potentially be used for vehicular ad hoc network testing. Vehicle ad hoc networks (VANETs) are an emerging technology planned to be implemented in new vehicles around the world - VANETs allow for increased security and safety while driving, and allows for greater access for information about entertainment and other interesting locations in the area. Despite this, privacy, security, and efficiency concerns are still prevalent, making VANETs a topic of interest in the research community.
Our testing process begins with the comparison of 3 realistic trace mobility models from the Generic Mobility Simulation Framework (GMSF) project. We created a simulator to test the data from the three scenarios: the urban, rural, and city models provided with the GMSF simulation, which obtains privacy metric data for each time step as the vehicles move across our simulation. The traces provided by the GMSF represent real vehicle movement pattens from areas in Switzerland. In addition, we use a mathematical Manhattan model as a means of comparison to the trace models in addition to providing us with a variety of scenarios. We take the metric data from tests of each of these scenarios and compare it to note any trends as mobility models change.
Patrick D’Errico, Warren Ma - Mobility Models' Effects On VANET Privacy
1. Abstract: A Vehicular Ad hoc Network (VANET) is a proposed network infrastructure that allows vehicles and infrastructure to communicate on an ad hoc basis, which is expected to be implemented globally in the near future. Such networks allow for communication between vehicles with themselves and vehicles and roadside units, allowing for significant applications, especially in terms of increasing safety for drivers. Privacy models have been put into place to test the privacy of these systems, a growing concern in them, and in order to properly and accurately measure the performance of these models, the topology of the roadmap must be considered - better road topologies may lead to better results regardless of the privacy models proposed. As such, we have developed a framework to analyze these models and track how they influence privacy in terms of three metrics: anonymity set size, entropy of the anonymity set, and tracking probability. Ultimately, while there are some differences between the models, density of vehicles has proven to be the major factor for change.
2. Our paper is in a draft state, with each section complete. We may do future work and merge with some of George’s work to develop a future paper to submit to conferences and journals. George Corser is our advisor.
3. Findings - While the mobility models do have some differences associated with them, density seems to be the overriding characteristic that drives the differences in metrics between the sets - higher densities provide a sort of diminishing returns in regards to these increases. We have also found inner city and urban models tend to give higher amounts of privacy in accord to the metrics used.
4. Contributions: a. within Discipline, b. to Other Disciplines, c.
to Human Resource Development, d. Beyond Science and Engineering
4a. We have developed a framework and means of comparison with which mobility models can be compared in terms of their effects on privacy metrics in simulations involving VANETs. We have also established a possible link between density and privacy.
4b. Through our initial study of attacker and privacy scenarios we have developed a comprehensive list of ways that privacy can be breached in the context of VANET systems. Through this, students of criminology or law can better examine possible future threats.
4c. Through our REU experience at Oakland University, we have learned much and grown as individuals. We have met many new individuals from diverse background, opening our minds to new ideas and ways of thinking. Under the guidance of George Corser and Dr. Huirong Fu, we have also gained invaluable research experience that will no doubt help us in pursuing other research experiences and applying to graduate school. We are very grateful for this having had this opportunity.
4d. We have learned about the research process in general and how to study an area which was initially unfamiliar to us. In addition, we have honed our presentation skills and developed the means in which to promote and pinpoint specific areas in which we have learned new things. Our research can also be used in law and engineering as well as our home field of computer science. Moreover, our specific field of study, VANETs, has already begun to appear in vehicles, so our experience has been applicable to the automotive industry.
5. As we worked on the “Mobility Models' Effects On VANET Simulations” project, we developed a well-rounded set of skills that will be of assistance in our future, whether that be in research or otherwise. We have learned about the construction and components of VANET systems and simulations that model them in addition to various privacy issues that generally affect networking challenges today. Furthermore, we have learned about python coding and simulation in general. Combined, these skills will surely be an asset in future endeavors.
We do this through simulation and analysis of several mobility models that could potentially be used for vehicular ad hoc network testing. Vehicle ad hoc networks (VANETs) are an emerging technology planned to be implemented in new vehicles around the world - VANETs allow for increased security and safety while driving, and allows for greater access for information about entertainment and other interesting locations in the area. Despite this, privacy, security, and efficiency concerns are still prevalent, making VANETs a topic of interest in the research community.
Our testing process begins with the comparison of 3 realistic trace mobility models from the Generic Mobility Simulation Framework (GMSF) project. We created a simulator to test the data from the three scenarios: the urban, rural, and city models provided with the GMSF simulation, which obtains privacy metric data for each time step as the vehicles move across our simulation. The traces provided by the GMSF represent real vehicle movement pattens from areas in Switzerland. In addition, we use a mathematical Manhattan model as a means of comparison to the trace models in addition to providing us with a variety of scenarios. We take the metric data from tests of each of these scenarios and compare it to note any trends as mobility models change.
Patrick D’Errico, Warren Ma - Mobility Models' Effects On VANET Privacy
1. Abstract: A Vehicular Ad hoc Network (VANET) is a proposed network infrastructure that allows vehicles and infrastructure to communicate on an ad hoc basis, which is expected to be implemented globally in the near future. Such networks allow for communication between vehicles with themselves and vehicles and roadside units, allowing for significant applications, especially in terms of increasing safety for drivers. Privacy models have been put into place to test the privacy of these systems, a growing concern in them, and in order to properly and accurately measure the performance of these models, the topology of the roadmap must be considered - better road topologies may lead to better results regardless of the privacy models proposed. As such, we have developed a framework to analyze these models and track how they influence privacy in terms of three metrics: anonymity set size, entropy of the anonymity set, and tracking probability. Ultimately, while there are some differences between the models, density of vehicles has proven to be the major factor for change.
2. Our paper is in a draft state, with each section complete. We may do future work and merge with some of George’s work to develop a future paper to submit to conferences and journals. George Corser is our advisor.
3. Findings - While the mobility models do have some differences associated with them, density seems to be the overriding characteristic that drives the differences in metrics between the sets - higher densities provide a sort of diminishing returns in regards to these increases. We have also found inner city and urban models tend to give higher amounts of privacy in accord to the metrics used.
4. Contributions: a. within Discipline, b. to Other Disciplines, c.
to Human Resource Development, d. Beyond Science and Engineering
4a. We have developed a framework and means of comparison with which mobility models can be compared in terms of their effects on privacy metrics in simulations involving VANETs. We have also established a possible link between density and privacy.
4b. Through our initial study of attacker and privacy scenarios we have developed a comprehensive list of ways that privacy can be breached in the context of VANET systems. Through this, students of criminology or law can better examine possible future threats.
4c. Through our REU experience at Oakland University, we have learned much and grown as individuals. We have met many new individuals from diverse background, opening our minds to new ideas and ways of thinking. Under the guidance of George Corser and Dr. Huirong Fu, we have also gained invaluable research experience that will no doubt help us in pursuing other research experiences and applying to graduate school. We are very grateful for this having had this opportunity.
4d. We have learned about the research process in general and how to study an area which was initially unfamiliar to us. In addition, we have honed our presentation skills and developed the means in which to promote and pinpoint specific areas in which we have learned new things. Our research can also be used in law and engineering as well as our home field of computer science. Moreover, our specific field of study, VANETs, has already begun to appear in vehicles, so our experience has been applicable to the automotive industry.
5. As we worked on the “Mobility Models' Effects On VANET Simulations” project, we developed a well-rounded set of skills that will be of assistance in our future, whether that be in research or otherwise. We have learned about the construction and components of VANET systems and simulations that model them in addition to various privacy issues that generally affect networking challenges today. Furthermore, we have learned about python coding and simulation in general. Combined, these skills will surely be an asset in future endeavors.