Towards a Proximal resource-based architecture to support augmented reality applications

We are developing a new enhanced cloud-based computing architecture, called the Proximal Workspace architecture to allow access and interaction between lightweight devices, e.g., video glasses, earphones, wrist displays, body sensors, etc., and applications that represent a new generation of computationand-data-intensive programs in areas such as augmented reality. While lightweight devices offer an easy way for these applications to collect user data and offer feedback, the applications cannot be run natively and completely on these devices because of high resource demands. Making these applications available via a cloud, while promoting ubiquitous access and providing the necessary resources to execute the applications, induces large delays due to network latency. To solve these problems, we are developing a new system architecture based on supporting workspaces which provide nearby computing power to the users devices and thus mediate between them and the clouds computing resources. Specifically, a workspace provides a set of middleware utilities designed to exploit local resources, and provide specific functions such as rendering of graphics, pre-fetching of data, and combining data from different servers. More generally, the workspace is designed to run any subset of activities that cannot be run on a user's device due to computation speed or storage size, and cannot be run on a cloud server due to network latency. Ultimately, the goal is to produce a set of middleware utilities that when run in the workspace with highly-interactive, computation/data intensive applications will result in better user-perceived performance. We are exploring this system architecture constructively, by adapting applications that benefit from this architecture and discovering how best to suit their needs. We have already adapted VNC (Virtual Network Computing, which allows local interaction with remote computations) to run under a similar architecture, and as a result increased video performance in high network latency conditions by an order of magnitude. We are currently working on adapting Google Earth to run under this system architecture, with the goal of the user being able to intuitively navigate through renderings of Ancient Rome in video glasses, without being hampered by any bulky equipment.