|
|
|
|
|
| Obviously, there is a great deal of work that need be done between now and when we have our ubiquitous computing environment. In an attempt to divide this problem up in to more manageable chunks, I've organized my thoughts into several key research directions: |
|
>
|
Bedrock (Distributed Security): Overcoming the obstacles of centralized security and administration models. |
|
>
|
Aquifer (Distributed Data): Creating a globally distributed repository of structured and unstructured data. |
|
>
|
MetaServer (Distributed Computing): Decoupling the software that is running from the hardware that runs it. |
|
|
|
|
Bedrock: Decentralized Security Model
|
|
|
|
| Existing centralized security and administration frameworks are inadequate for managing such a complex, amorphous structure as our ubiquitous computing framework will likely be. Bedrock provides a general-purpose, extensible security model that suffers from no centralized development, deployment, or decision-making bottlenecks. |
|
|
|
|
Aquifer: Globally Distributed Data Repository
|
|
|
|
| While the Web does a great deal to unify the manner by which data is stored and distributed, it is inherently limited by its fundamental assumptions. Most notably, assuming the presence of a fast, reliable network between any two points greatly limits the Web's reach and robustness. Likewise, the Web's event-driven, copy-centric nature complicates the design and deployment of highly distributed, shared-data applications. Aquifer is an evolution of existing Web concepts designed with an eye toward the needs of ubiquitous computing. |
|
|
|
|
MetaServer: Realtime Distributed Supercomputing
|
|
|
|
| The promise of Web services, and the other Remote Procedure Call (RPC) technologies before it, should not be taken lightly. The elusive goal of reusable, interchangeable components is well worth the extraordinary effort put forth to build these technologies. However, I believe these technologies fail to realize their potential due to defining the problems they solve to narrowly. By expanding the purview of the RPC technologies to include the newest component-container and agent-based concepts, a truly distributed computing network can be formed. This network, called the MetaServer, can once and for all break the traditional ties between the physical limitations of hardware, and the digital strivings of software. |
|
|
|
| Current systems and applications generally make the assumption that they "own" their users. This is evinced by the expectation that users must create new accounts, with new names and new passwords, at every Website they visit. The presumption made by this technology is that these user accounts are separate and distinct -- that the data in that account is "owned" by the Website that does the gathering. Talisman is a new system, similar in intent to Microsoft Passport (but diametrically opposed in design), that returns ownership of accounts to users, enabling all the privacy and maintenance benefits afforded. |
|
[40 children...] |
|
|