The Materials Microcharacterization Collaboratory

Participants

Objective

Approach

Accomplishments

Plans

Tools Available

Publications

Demonstrations

Partners

Overview

Now that we have a number of instruments online to varying degrees, we plan to increase our emphasis on use of these instruments for actual materials science experiments.   We are migrating some of our systems from prototypes to a more production oriented environment to make them easier to use and more reliable.  We are creating new web sites to make our facilities and capabilities easier to find and understand.  Our many demonstrations have created substantial awareness or remote collaboration within the scientific community and many people are interested, in principle, in remotely collaborating with us.  We are planning to make that easier to do.  We plan to implement an online scheduling and session control system.

The educational community has shown a strong interest in our project and we are pursuing several initiatives to secure additional funding to expand the MMC into the educational arena.

Of course, we are by no means finished with development of collaboration technology.   Work continues at LBNL on the uniform architecture and we have begun porting it to the ORNL SHaRE facility.  We plan soon to evaluate it for the beamline instruments as well.  We also plan to soon provide at least a limited web interfaces for the instruments that do not yet have a web interface.  We will convert some video streams to higher quality tools.

Work is still needed on remote control capability for specific instruments.  For example, the Windows NT version of the control system for the Philips XL30 is expected soon.  The new motion control hardware for the neutron beamline has arrived but not yet been installed.  The X-ray beamline at BNL has been moved from a VMS-based to a Linux-based control system but remote operation has not yet been implemented.    We plan to expand remote control to other instrument already at our facilities and will specify remote capabilities for instruments being purchased (ex, state of the art FEG EF-TEM/STEM).

The experience we gain through increased usage of collaboration tools for actual collaboration will be fed back into the continued development of those tools.  After, that is what a pilot project is for.  We have been making significant use of the ORNL version of the DOE2000 electronic notebook and have several improvements to suggest, particularly in terms of user-friendliness.  Some of these capabilities already exist in the PNNL version of the notebook and we will take another look at that.  We are also providing feedback to commercial tool developers (ex, GTS, IPIX, National Instruments).

Plans for Uniform Architecture Development

Before the end of the current fiscal year, we plan to

• Incorporate an insitu computational service for HVEM
• Integrate the secure socket level (SSL) as defined by OMG
• Explore and incorporate a DOE2000 security component
• Integrate the DOE2000 notebook to our baseline system.

Future work will include

• Addition of new microscopes and/or functionalities
• Addition of new computational services
• Performance evaluation and measurement
• Addition of Quality of Service (QoS) and bandwidth reservation.

We will add new instruments into the current microscopy channel, and integrate the baseline system with vendor's analytical tools. The first objective will be met by applying our framework to either new microscopes or current on-line microscopes in different modalities. Let's clarify the later point. The current implementation of on-line microscopy for CM300 is limited to high resolution transmission electron microscopy. However, this instrument can also operate in EELs as well as Holographic mode. Furthermore, current generation EMs are not designed for remote operation. As a result major efforts have been invested on HVEM and CM300 for retrofitting. Additional investment will be required.

The second objective requires developing a COM interface to vendors analytical software tools. Microscopists use these tools (through scripting languages such as JavaScript) for either data processing or placing the instrument in a particular state for image acquisition. A vast array of analytical capabilities exist within vendor's domains that are actively being used by microscopists. These utilities should be brought on-line for full functionalities.

We plan to integrate Berkeley's capability in image simulation to our baseline system.

To a large degree, the design of IDLs has been influenced by writing test programs to evaluate the behavior of bulk data transfer over the high speed network (100Mb/sec). All comparisons were against direct socket with TCP transport. The results indicated that Iona's ORB middleware is about 7.5% slower, and the Event Channel transmits information at the rate of the client with the least amount of bandwidth. However, some of these factors are about to change with the new release of middleware that is faster and more efficient. Nevertheless, the behavior of the Event Channel will require a new design for accommodating the low bandwidth clients. Furthermore, no performance measures for multicast sessions were collected.

We will use elements of TAO (University of Washington) to ensure end-to-end QoS over the high speed network. Furthermore, we will integrate the network bandwidth reservation component that is currently being developed as a part of DOE2000 technology component.

Plans for Security Infrastructure

Now that we have microscopes and beamlines actually online, we can start deploying a more general security architecture. There are several issues to consider here

• While the MMC does need to maintain an archive of shared stored data, most user data is generated and transferred to the user's own computer system.
• Electronic notebooks have been our main vehicle for sharing information, and a combination of specialized notebooks, encryption, and strong access control seem to suffice from a security point of view.
• Our users usually are performing one-time experiments.

Thus, we plan to issue short-lifetime certificates to our users and guests. If a certificate expires before an experiment ends, it can be easily renewed. A short-lifetime certificate eliminates the problem of certificate revocation lists.

By examining the status of the user (in the DN), we can filter the programs that a user is allowed to access. By further implementing the Akenti mechanism, we can enforce stakeholder rights if necessary.

In addition, we plan to put certificate-based security hooks in the CORBA architecture used in our LBNL middleware.