|
|
|
|
CISE - CNS
COMPUTER SYSTEMS RESEARCH (CSR)
The CSR program seeks proposals focused on compelling research problems with potential to advance the
state-of-the-art on how to improve the design, use, behavior and stability of computer and software
systems within and across the following systems areas: distributed and Internet scale computing, massively
parallel and data intensive computing, and pervasive and ubiquitous computing.
Broad categories of research interest within CSR include, but are not limited to:
- Scalable and robust frameworks, methodologies, algorithms and tools for resource management,
protection and virtualization in large-scale heterogeneous, potentially failure prone,
environments,
- New directions and innovative approaches for the design, implementation and management of modern
storage and file systems, including energy-efficient and self-managing storage, access-anywhere and
personal storage,
- Innovative approaches to massive data and metadata access, caching, replication and
consistency,
- Scalable, context-aware and energy-efficient system services,
- Scalable and robust system and software architectures, models and programming abstractions to
support changing trends and emerging technologies, such as sizes or speeds of processors, access
memory and storage, data-intensive computing, ubiquitous and pervasive computing and peer-to-peer
computing,
- Energy and context aware paradigms, methodologies and tools to improve system manageability,
configurability, operational sustainability and performance, and to reduce vulnerabilities while
improving usability,
- Frameworks, methodologies and tools for quantitative and qualitative evaluation, monitoring and
prediction of complex computer systems behavior and performance.
- Fundamental and system-level research on energy-aware architectures and design
methodologies,
- Novel parallel programming models and abstractions, compiler and dynamic run-time support for
parallel programming and coordination languages, and advanced resource management frameworks,
methods and tools to support highly parallel data-intensive computing environments,
- Innovative energy-efficient, fault-tolerant run-time execution environments, service architectures
and coordination frameworks to enable large-scale concurrent execution across heterogeneous parallel
and distributed computational platforms,
- Advanced frameworks, programming abstractions and models for parallel computing
environments,
- New paradigms, frameworks and tools for automatic parallelization, synchronization and
concurrency control,
- Power and energy-aware compilation and runtime optimization techniques for parallel computing,
including dynamic and adaptive compilation, automatic code generation, program characterization and
phase analysis techniques for optimized performance, and computation steering for reliability,
scalability and improved performance,
- Advanced resource management frameworks, methods and tools, including scalable scheduling
algorithms for resource and data intensive parallel systems, multi-criteria scheduling frameworks
and algorithms, tools and environments for workflow scheduling in parallel systems, and adaptive and
dynamic load balancing algorithms and tools,
- Application and system level methodologies and tools that exploit the characteristics of the
hardware and execution environment to achieve high-level parallelism; and novel frameworks,
methodologies and tools for performance prediction and evaluation of complex parallel systems and
large-scale parallel applications,
- New paradigms, methodologies, algorithms and tools to enable robust and highly reliable real-time
systems, across diverse computing and software platforms, capable of operating in widely distributed
and highly interactive and uncertain environments.
- Novel frameworks, methodologies and software systems for distributed sensing and data
- New paradigms, mechanisms and tools for real-time resource management that address and integrates
multiple resource constraints and performance requirements, such as power, clock frequency and thermal
gain, task dependence, real-time guarantees, and criticality level.
|
|
|