Georgetown University proposes to establish a PKI Lab as a project of the Georgetown Institute for Information Assurance (GIIA).  The mission of the Institute is to provide leading edge research and education in information assurance and related disciplines, including information security, privacy, information warfare, computer crime and forensics, intellectual property protection, fraud prevention and detection, and strategic communications.  It seeks to advance the practice of information assurance, including policy, social, ethical, legal, and technology issues.  GIIA is the joining of several areas of the University:

• Georgetown Main Campus, which includes the Computer Science Department. Dorothy Denning, Professor of Computer Science, is Director of the GIIA.
• Georgetown University Medical Center (GUMC). GUMC is the medical school and maintains very close operational relationships with the Georgetown University Hospital. It is actively involved in several health care security and privacy projects, including projects related to HIPAA.  Participants include Jeff Collmann, Associate Professor of Radiology and Director of the Health Information Security Program; Janlori Goldman, Director of the Health Privacy Project; and Alan Zuckerman, Assistant Professor of Medical Informatics, Department of Family Medicine.
• Georgetown University Law Center.  Law professors are participating in the legal aspects of information security through established national projects on social and legal implications of security and privacy for health care.  Participants include Lawrence Gostin, Professor of Law and Principal Investigator of the Model State Public Health Privacy Project: Privacy and Security of Public Health Information; and James Hodge, Adjunct Professor of Law and Director of the project.
• University Information Services (UIS), the central IT organization wishing to deploy a PKI on top of the existing directory based (LDAP) services.  Participants include David Lambert, Vice President and Chief Information Officer; Michael Gettes, Lead Applications Systems Integrator; Richard Kogut; University Information Security Officer and Chief Information Technology Architect; and Brian Reilly, Security Agent.

Georgetown University has demonstrated expertise in the issues of implementation and deployment of directories and directory-based services.  This expertise has been utilized within the Internet2 community on projects within the Internet2 Middleware initiative as well as working with other Universities to develop and teach methodologies for directory services.  We have been an integral part of development in the eduPerson objectclass effort for higher education and the higher education directory of directories project.  Directories are an integral aspect of PKI that is often overlooked and, at times, over-simplified.

Several of the principals who will be participating in the PKI Labs project bring strong theoretical and practical experience with information security to the development and deployment of a PKI.  Dorothy Denning is well known for her research and expertise in cryptography and information security, including her work on key management and key recovery protocols and technologies.  Jeff Collmann, a social anthropologist, has focused his research on how sound security practices become incorporated into routine organizational work practices.

Michael Gettes, Lead Application Systems Integrator within UIS has developed and deployed server-based PKI at both Princeton and Georgetown Universities.  Both deployments involved extensive directory infrastructures that formed the foundation for further development of client oriented PKI.  Financial costs and lack of standards have prevented the full PKI deployments at both institutions thus far.  Development has always been oriented towards practices that can be leveraged at other institutions. Georgetown aims to be a "model campus" for IT infrastructure and central services.

Alan Zuckerman is experimenting with biometrics technologies (fingerprint  and iris recognition ) as a proof of concept for application integrated software.  The approach uses the Java Native Interface (JNI) to integrate the native C language DLLs running under Windows 95/98 into server-based authentication systems.  The biometrics  records are stored in SQL databases or LDAP directories using encryption of transmission and data records.  Zuckerman is also using the Sun Java Cryptography Extensions to demonstrate key features of digital signatures and explore integration into a campus-wide PKI.

The Federal Bridge Certificate Authority model, while not widely accepted or known, is the logical framework for the future of PKI.  To that end, our work would be centered on a BCA model.

Our initial effort will include the following:
 

1. The General Issues section of the RFP lists the obvious set of issues facing PKI today.  CRLs do not appear to be practical for large scale PKI, especially for the high turnover rate of communities at a University.  OCSP seems to be a reasonable alternative where CRLs will not scale.  We will consider extending OCSP for path construction and validation in a BCA world.  If OCSP is not well suited, then the construction of a separate path construction and analysis protocol and server will be necessary.  This will allow for minimal knowledge in client deployments.
2. Like most other universities, we are seeking a cost-effective solution toward the deployment of a PKI.  Given the prohibitive costs involved in vendor CA solutions, and vendor outsourcing, we intend to investigate the Open Source arenas and leverage having the source code to develop missing or immature components on the CA, web server and client fronts, initially.  This will allow us to develop an application framework that would allow anyone, vendors included, to easily incorporate PKI into existing applications.
3. In the authorization area, attribute certificates vs. identity certificates are certainly areas worth exploring, but further work needs to be done to explore the pros and cons of directory based attribute certificates; that is certificates that have pointers to directories that contain the attributes associated with the certificate as opposed to loading the attributes into the certificate with extensions.
4. We will explore the use of portable keys with browsers by developing suitable protocols and encryption techniques that would enable the secure stowage of this material on central servers and accessible by the two major browsers.  This may be an alternative solution towards bulk deployment of certificates.  This work would be done in concert with HEPKI-TAG, the Higher Education PKI Technical Advisory Group.
5. We will explore approaches to storing and protecting users' private encryption and signature keys.  Central storage of private keys would facilitate access across campus, but could threaten privacy and non-repudiation of signatures if users do not have control of their keys.  Encryption  coupled with strong authentication, perhaps based on biometrics, might be used to control access to the private keys, whether they are stored centrally or on tokens or smart cards.
6. In an academic environment, faculty and students are not likely to want their private keys held by the university for the purpose of key recovery.  We will develop an approach to recovery that takes this into account as well as the needs of the university to ensure access to university-owned information.
7. We will engineer the Apache Web server to interoperate in a BCA model.
8. We will conduct vulnerability assessments of the approaches we take, seeking methods that address such threats as unauthorized access, denial-of-service, tampering, spoofing, and malicious code.
9. We will develop a model system for coordinating Human Resources data with PKI registration in a manner that is supportive of institutuional policy about individual access and use of information resources.
10. The PKI needs to support secure e-mail with patients who are not regular members of the Georgetown community.  We will develop a mechanism for issuing patient certificates that allows physicians to communicate securely with their patients.
11. Ultimately, the PKI must interoperate across all  area medical schools participating in the Area Health Education Center (AHEC), which will have medical and nursing students rotating to a network of community-based clinics.  The Georgetown PKI will create a testbed to coordinate with other universities so that their students  can access health science resources at Georgetown using a trusted bridge between the schools.  We can simulate a small PKI for a test group of students at the other universities until they are ready to implement their own PKI.
12. We will examine what kinds of user support are necessary to make PKI effective in a university environment.  We will evaluate different approaches to user training in the use of PKI.