SBIR-STTR Award

A distributed computing environment (DCE) provides seamless, locati on-transparent computation atop heterogeneous hardware and software so that DCE resources are accessible without knowing their location. However, some technical approaches to location-transparency can be problematic for mobile computers, such as "notebooks", "laptops", etc., because certain CED technology may be intended primarily for desktop or workstation systems with relatively fixed locations. One potential problem is inflexible security mechanisms designed for fixed-location systems. Also, when a complex DCE requires decentralized administration and must be divided into multiple "clusters", other mobile computing problems can arise; such as name ambiguity where the same user name or object name could refer to different things in different parts of a large DCE. Furthermore, although portable computers are rapidly increasing in performance and capability, we cannot ignore user needs for low-cost devices with limited capabilities that may not be able to directly implement highly sophisticated DCE software. Our research will define a functional design baseline for DCE portable computing that addresses these problems. To the extent possible, our approach is independent of any particular DCE technology to maximize dual-use.

Keywords:
Portable Mobile Distributed Computer Cluster Heterogeneous Secure Communications

Mobile computers have obvious benefits to users of distributed systems. Aside from the mere convenience of mobile system access, survivability can be enhanced by moving to alternate network points following failures. Yet, in order to realize mobility benefits, a great deal of effort must go into proper system architecture and design. Many approaches to mobile distributed computing focus on the "mechanics of mobility"; the basic protocols needed for dynamically chantging node position. Such approaches are highly limited in scope and can lead to systemic architectural inefficiencies. Recent research has produced an innovative Mobile Distributed Computing Architecture (MDCA) that uses migratory Session-Layer services that continually "wander near" the portable systems for which they provide service. Theoretical analysis indicates that the MDCA can retain the reliability, availability, and survivability expected of distributed systems, while imposing two to three times less communication load on "backbone" networks versus more conventional approaches to mobile distributed computing. Our current aim is to build a prototype of the MDXA to demonstrate its features

Keywords:
Mobile Computer Portable Migration Distributed Heterogeneous Survivability Communications