Hi. This is my official web page. I am:

• James E. Marca
• Postgraduate Researcher
• Institute of Transportation Studies
• 545 Social Science Tower
• University of California
• Irvine, CA 92697-3600
• jmarca@translab.its.uci.edu

I am a post-graduate researcher in the Institute of Transportation Studies here at UC Irvine. My research focus for the past several years is on using wireless technology to improve data collection techniques in particular, and distributed traffic control in general.

Some papers that I've written or collaborated on are listed below

• The personal travel assistant (PTA) project. We recently submitted this project as a proposal to the NSF. Hopefully it will hit, but if not, it is still a good idea and we'll get funding somewhere. In essense this project continues to push all of the themes I have been working on for the past 6 years or so: automatic data collection; learning travel and activity behavior without pestering the traveler with a million questions; providing useful travel information to a traveler; and recognizing that travel information is essentially worthless in the end. The proposal is cool because it builds from practical to theoretical, first designing a data collection system, then using that system to perform some basic microeconomic research, as well as to push the boundaries of traveler information systems research. One way or the other I'll post more information on this project by mid-summer, probably with a draft TRB paper.
• Modeling Travel and Activity Routines using Hybrid Dynamic Mixed Networks (PDF), presented at the 85^th annual meeting of the TRB. This paper is the description aimed at a transportation audience of research we are conducting with Rina Dechter's group here at UCI trying to apply Bayesian Belief Networks to learning travel and activity routines. The idea is pretty simple: we suppose that people have habits and patterns of behavior, and therefore we can learn those patterns of behavior by studying a person's travel history. In this case, the travel history is collected using GPS data. In fact it is the same data that I used for my dissertation research, except in this case Vibhav is processing the trips between destinations, not just the destinations. Vibhav has written more technical papers for a computer science audience. Craig and I revised this paper to present the basic idea and it applications. This work forms the basis of our activity model in our PTA project
• A Method for Creating a Real-Time, Distributed Travel History Database: The PTC Project (PDF) Draft paper, presented at the 85^th annual meeting of the TRB. This paper is slightly different version to one that will be published in an upcoming issue of TRB's Transportation Research Record. This work was supported in part by NSF grant #0339307
• Mobile throughput of 802.11b from a moving vehicle to a roadside access point (PDF). This paper, also presented at the 85^th annual meeting of the TRB, documents the observed characteristics of communications over 802.11b from a vehicle to a roadside antenna. The main flaw in the paper is that I can't seem to get the signal to drop to zero due to speed or distance effects. Given the geography, it is impossible to get far enough away without dropping behind a hill, and it is impossible to go fast enough while in range without breaking the law. That's pretty good news for 802.11b, I guess.
• Design, implementation, and test of wireless, peer-to-peer, roadway incident exchange pdf
  written with Trevor Harmon, Ray Klefstad, and Peter Martini, this paper is the first of hopefully many papers that we will write on the Autonet and related concepts. This paper will be presented at the ASCE AATT conference this summer (2006). Trevor, Peter, Ray and the Ray's DOC gang did the in-vehicle device, wireless protocols, and ran the tests, etc.
• Agent based transportation modeling (ps) I presented this paper at TRB 2000. It describes our approach to agent-based modeling of activities and travel patterns. My co-author, Dr. Craig Rindt, also believes that it is the first paper to argue that human activities can be modeled as the outcome of a negotiation process between independent agents, where the agents are not just "fake people" but also land use resources, the travel network, etc.
• On-line travel and activity survey (pdf) This paper won the best stuent paper award at the ASCE Applications of Advanced Technology in Transportation (AATT) conference in 2002. It is short due to the requirements of the conference, but it succintly describes the lightweight travel and activity survey I designed and implemented as part of my dissertation work. This survey is somewhat unique in that it generates questions on the fly based on a real-time stream of GPS data coming in from the participants' vehicles. It could be used to generate in-vehicle surveys, since the time lag is only a few seconds between when the data is received from the GPS recorder, and when it has been safely tucked away in the database ready for use.
• Using spatial krigging to infer activity purposes (pdf) One of the tools I developed to process the collected GPS data was a set of R programs that applied spatial krigging to the data. The idea is that people will quickly tire of entering their activity information for every destination. However, using even a small number of answers, it is possible to guess approximate activities for a large number of recorded destinations. A good example is going to work. While the GPS coordinate of your parking spot is going to be different every day, you will get pretty fed up with saying "work" to an on-line survey that is asking you what you did at some location. By using kriging techniques, one can properly account for small spatial variations, and then estimate the likelihood that a particular activity occurred at a particular, as-yet-unlabeled destination. Furthermore, kriging points up destinations about which no guesses can be made, and destinations which are equally likely across a few activity types. Asking about these questions is a good use of the respondent's time, while asking about well known destinations will probably annoy the respondent.

Here are some other links.

• Academic, a website I tinker with to handle on-line research. Ah, it is dead now. Sorry. Here is the source code, provided as is under the GPL version 2, with no warranty. That site is a plug-in of sorts for Slashcode. That is both how I got it up in the first place (I bootstrapped my Perl and MySQL knowledge studying Slash) and why it is no longer up (I no longer run slashcode on any of my computers).
• The Tracer website, a web site set up for accessing Extensible Data Collection Unit (EDCU) data (GPS points). I developed this website in Java using PostgreSQL/PostGIS, Struts, Torque, and Velocity, and running on Tomcat. It was based on an earlier system using Perl and Slash running on Apache. Access to the site is controlled by the UCI Testbed system. If you have an account you can access the site. If you don't, the information page may be enlightening.
• The Tracer website is mostly designed around handling the data streaming in from our EDCUs. The EDCU is a Linux box (originally the LRP system, but now more or less a LEAF system (LRP's successor) running on a PC/104 board with 32MB ram, a 16MB CompactFlash card, a GPS antenna, and a CDPD modem, which worked great until it became obsolete. The PTA project, described above, will be the successor to this project when it comes online. We have 45 EDCUs, and use them to collect real-time travel information. Here is an older website desribing the box.
• Most of my code is freely available under the GPL version 2, or the Perl Artistic License, depending on whether or not I changed the relevant bits in the autogenerated .pm files.
• My family. Mostly my daughters, Grace and Emma. Pictures of Grace and Emma, and sometimes Brooke and me
• My Grace-Notes code is a set of perl modules designed to run under Catalyst. I've been developing it for the past 6 years, since the birth of my oldest daughter. I recently converted it to run under Catalyst instead of Slash. It allows one to manipulate and annotate a database full of images, for the purpose of rendering static "chapters" of related images. That's how we generate our website. Right now I think I've got just under 19,000 original images in my database. We'll probably add another couple thousand by the end of the summer, no doubt. Here is the latest version of the code, a snapshot as of July 27, 2006: GraceNotes_07272006.tgz. It is close to running out of the box, but not quite. There are a few database entries that still need to be made by hand first. Also be warned that you have to install lots of stuff through CPAN in order to run this. A good start is to install Catalyst and DBIx::Class and then see what classes are missing. Someday I'll figure out how to make a pre-requisites file and keep it up to date.

If you want more details, like my resume, more detail on my research, or to collaborate on a project, email me.