Distributed and Adaptive Processing
We have developed and demonstrated several distributed-processing technologies
that extend commercial database management systems by using weak-consistency
methods to automatically replicate partial copies of a network database.
We have applied one of these technologies to a geographically distributed
database of weather information.
Other SRI technologies use data flow concepts to enable applications
to execute in parallel on multiple workstations in a network. We have
examined several applications of these technologies in areas including
AI, parallel make, and ray-tracing applications. We are also developing
load-balancing techniques for use in coarse-grained parallel processing
and distributed real-time systems.
Distributed Interactive Simulation
We have
developed DIS applications for command, control, communications, and intelligence (C3I)
that incorporate simulated sensor and intelligence processing. We also have
investigated methods to integrate simulations with deployed, automated
command and control (C2) systems. Protocol Suites for Distributed Processing
In order to address the shortcomings of many commercially sold Remote
Procedure Call (RPC) and Interprocess Communication (IPC) protocols, SRI
developed a protocol suite for procedural IPC (PIPC). These protocols
support the standard model of RPC, and also provide asynchronous and message-based
calls to support efficient distributed processing. A second suite, the
Point of Contact protocols, provides asynchronous and dynamic management
of input buffers, allowing applications to continue processing while acquiring
data, and thus increasing the parallelism (and improving the performance)
of networked and distributed systems.
|
R&D Divisions