The Secure Document Systems area at the Xerox Palo Alto Research Center (PARC) is contributing to the advancement of digital security in several key ways. These include:

• Fostering and participating in the growing movement towards abundant and pervasive use of cryptography
• Making key contributions in the areas of cryptosystem performance and usability
• Conceptualizing services, protocols and businesses that will be made possible by abundant cryptography

The Quicksilver project is an umbrella that encompasses a number of research areas. Some of these areas are themes that we pick up from time to time while others are day-to-day projects. The work may cross more than one research area or it can be independent. We also do research in non-cryptographic infrastructures to enhance network security.

With cryptography becoming cheaper and cheaper, what kind of infrastructure can we provide to enable secure networks? Work in this area includes offering fast cryptography as a network service and building systems for fast and secure group communication. ParKI is our project to specify and develop public key infrastructures at PARC.

In particular, we have built a high-performance cryptography server able to deliver 1024-bit RSA operations at a cost of less than $0.000001 cents each over the Internet. Such a device has many possible applications including terminating SSL connections, micropayments, and any other use of the RSA algorithm.

In the near future, there may be many fast cryptoservers on the network. There are several scenarios under which it is natural for a group of cryptoservers to compete. We identify a number of mechanisms that will help make competition among cryptoservers more effective: variable pricing, recommendation systems, and auditing. There are special advantages to implementing these mechanisms for cryptoservers as opposed to any other (generic) commoditized service.

We envision a future in which wireless home-area or even personal-area networks will be abundant. (Imagine your cell phone talking wirelessly to your PDA and your ear piece.) We are investigating how these networks can be secure while, at the same time, permitting ease-of-use and intuitive security management.

Analyzing the security of existing systems and pointing out vulnerabilities allows us to both educate the public about security threats and to investigate good and bad security practices.

There are a number of security protocols for which the degree of privacy scales nearly linearly with the computational burden. We propose a new business model for secure services where a protocol participant pays for computations in a privacy-scaling manner. We can think of this as hiding the participant's true intentions or true identity among a forest of alternatives, and charging the participant "per tree."

There are many applications where one desires data to only be available in a useable form after some sequence of actions has occurred. In a workflow application, we might only give the purchasing agent access to the purchase requisition after the appropriate digital signatures have been applied. Modern computer simulations (e.g., games) come with a simulated, but rich environment. Particularly for multi-player games, the producer of the environment might wish to ensure that a participant (player) has only seen the subset of the environment that corresponds to states they have legitimately entered (i.e., parts of the game that they have played). Military maps could benefit from a similar scheme. Encryption allows the data to be distributed (e.g., via DVD-ROM) to users, to allow for low latency access, while maintaining the secrecy of the data.

It is impossible at present to determine where a network packet originated or what path it took through a network. Route-tracing information associated with packets would facilitate many improvements in data communication including billing for network service, routing based on policy, verification of quality-of-service guarantees, and prevention and investigation of security problems. We envision several methods to associate routing information with packets. One of our ideas, the pile-on authenticator, has the advantages of being a fixed size no matter how many hops the packet takes and of requiring no storage at network routers. This approach is cryptographically heavy but becomes practical given the steep drop in the cost of public-key cryptographic operations expected over the next several years.