Site visit and report completed by Mark Gandolfo and Greg Gardella

General Observations

The quality or lack of quality, of the Davis node particularly, magnified the inherent shortcomings of the larger Access Grid Node concept. Overall, the AGN model is flawed from a communications perspective. While it is obvious that great effort and thought have been applied to the delivery methods, a basic lack of understanding of fundamental visual communications principles have undermined the true collaborative intent of these projects. Some of these problems can be mitigated by having an experienced operator in the room, but this would require a very focused effort over the duration of a meeting, which might not be feasible.

It is hard to determine which site is speaking due to the number of video streams per site (3 to 4) multiplied by the number of sites participating. The resulting large number of projected windows is confusing and disorienting. Further contributing to this is the low-resolution, highly-compressed video which does not transmit enough detail to easily determine which person is speaking in which window. The problem is somewhat reduced if the operator zooms in on the specific speaker. Though the lack of audio sync makes this difficult if not nauseating to watch

The video window matrix layout is arbitrary necessitating an operator arrange the windows in a logical manner to reduce confusion.

Placement of cameras varies between sites, and, in general, sites as observed do not convey an understanding of environmental context. It is very difficult to determine a remote site’s room layout from the illogical 3 to 4 video streams they are transmitting.

Video quality varies greatly between sites further contributing to an underwhelming communication experience. Better sites have specific professional video lighting and proper backgrounds, i.e. middle value grays or blues.

Background -- Davis

The U.C. Davis Access Grid Node was produced as a part of a grant originating outside of the University, but necessary to fulfillment of the grant. It was installed approximately 13 months ago and was a turnkey system purchased directly from Insors Integrated Communications, Inc. The system came to them as a populated rack and two campus computer network administrators were responsible for the setup. The installation, which was not completed by Insors or MCSI, went smoothly. The network connectivity posed a series of problems which ultimately took 2 months to resolve. These issues related to interoperability of their network routers when the AG node was multi-casting. Apparently, they had a mix of Foundry, Juniper and Cisco routing hardware and this mix had issues with the InSors setup. This resulted in their being filtered from I2 until the problems were rectified. The following list serves as a group of observations which are descriptive of this site, specifically and in some cases of Access Grid Nodes in general.

Background – Berkeley

The U.C. Berkeley Access Grid Node was produced as an internal initiative at Berkeley, through the computer science department there. This node was planned, designed and built by Berkeley personnel. While the configuration follows general recommendations made by the AG organization, this room should be considered a hybrid of the traditional AG node. A number of enhancements, which broaden the capability of the node and make it easier to use, have been incorporated into the configuration. These enhancements will be listed below.