SARANA: language, compiler and run-time system support for spatially aware and resource-aware mobile computing

Area of Science:

• Computer Science
• Distributed Systems
• Mobile Computing

Background:

• Spatial awareness is crucial for modern distributed and mobile computing.
• Existing systems lack efficient programming models for spatially aware applications with dynamic constraints.
• Programming these applications requires addressing geographical position, service information, and real-world platform dynamics.

Purpose of the Study:

• Propose the SARANA system architecture for spatially aware and resource-aware applications.
• Provide language and run-time support for expressing spatial constraints and quality of result (QoR) trade-offs.
• Enable efficient application execution on resource-constrained, dynamic platforms.

Main Methods:

• Developed the SARANA system architecture with integrated language and run-time support.
• Implemented a system allowing users to define regions of interest and balance QoR, latency, and cost.
• Designed a run-time system for dynamic management of QoR and cost trade-offs via resource tracking and program migration.
• Incorporated a resource cost model for user-defined resource needs and QoR expectations.

Main Results:

• SARANA allows expression of spatial regions and trade-offs between QoR, latency, and cost.
• The system dynamically manages QoR and cost by tracking resources, brokering agreements, and migrating programs.
• A resource cost model facilitates user expression of needs and expectations.
• Initial demonstrations on a nine-node prototype show system viability.

Conclusions:

• SARANA offers a novel architecture for developing efficient, spatially aware applications.
• The system effectively manages resource constraints and dynamic environments for mobile and distributed computing.
• Further development and testing are ongoing, with promising initial results.