Dataflow-Based Implementation of Layered Sensing Applications
Abstract
This report describes a new dataflow-based technology and associated design tools for high-productivity design, analysis, and optimization of layered sensing software for signal processing systems. Our approach provides novel capabilities, based on the principles of task-level dataflow analysis, for exploring and optimizing interactions across application behavior; operational context; high performance embedded processing platforms, and implementation constraints. Particularly, we introduce and deliver novel software tools, called the targeted dataflow interchange format (TDIF) and Dataflow Interchange Format Markup Language (DIFML), for design and implementation of layered sensing and signal processing systems. The TDIF-CUDA (Compute Unified Device Architecture) environment is a graphics processing unit targeted software synthesis tool that provides a unique integration of dynamic dataflow modeling; retargetable actor construction; software synthesis; and instrumentation-based schedule evaluation and tuning. The DIFML package is a software package for the DIFML format, which is an Extensible Markup Language (XML)-based format for exchanging information between DIF and other tools.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2011
- Accession Number
- ADA540319
Entities
People
- Chung-ching Shen
- Hsiang-huang Wu
- Nimish Sane
- Ruirui Gu
- Shuvra Bhattacharyya
- William Plishker
Organizations
- University of Maryland