Computer Systems Research (CSR)

Computer systems support a broad range of applications and technologies that seamlessly integrate into a cohesive, intelligent computational ecosystem driving scientific progress, economic prosperity, and national security. Revolutionary computing capabilities underpin the emergence and dramatic growth in areas such as intelligent systems, biotechnology (e.g., drug discovery, genomics, and personalized medicine), data science and analytics, smart homes and transportation, advanced manufacturing, quantum computing, and virtual/augmented reality. The challenge ahead is envisioning new approaches, as well as combining existing and emerging technologies, to create the future computer systems enabling continued transformational growth. Such computer systems will require new, innovative, and visionary holistic approaches to integrating hardware and software systems, comprised of classical and quantum computing elements, networking and communications, and memory and storage components, into computer systems spanning from the largest hyperscaled systems and supercomputers to ubiquitous IoT and smart devices embedded everywhere. The scope of the Computer Systems Research (CSR) program includes the integrated hardware and software of classical computing systems, quantum computing systems, embedded and constrained computing systems, as well as other computing systems such as analog, asynchronous, or neuromorphic systems.

Classical Computing Systems

Over the past 75 years, computing systems based on the classical, von Neumann architecture have transformed all aspects of society, fueling tremendous research and development breakthroughs, revolutionizing productivity throughout the economy, and enabling technological superiority in the national defense. Next-generation computer systems must tackle grander challenges, support wider deployments and exponential growth in data and processing requirements. These systems will need to be reliable in the presence of unreliable components, networking, and configurations, adaptive to changing environments and usage scenarios, attentive to unintended consequences of their operations, capable of supporting high-throughput applications and large-scale data storage and processing, maintain privacy and security, and able to meet performance and energy objectives for applications ranging from very low-power embedded systems to large high-performance computing systems. In addition to hardware improvements, the computer systems software stack must continue to provide the programming languages and models, compilers, libraries, middleware, and application interfaces needed for developers along with the operating and runtime systems, virtualization, resource management, and orchestration support required to support the full range of deployed computer systems. Moreover, the networking and interconnection technologies employed for integrated and distributed computing systems continue rapid technological change that demands judicious approaches to creating balanced, performant systems. Similarly, memory hierarchy and storage systems must support disparate requirements for temporary and persistent storage at scales ranging from a handful of bytes to internet-scale data training sets. Holistically integrating and balancing these aspects of computer systems remains a tremendously exciting research endeavor with critical impact on computing capabilities. Consequently, the CSR program seeks innovative research proposals that will significantly advance the reliability, performance, power, energy-efficiency, security and privacy, scalability, and sustainability of computer systems.

Quantum Computing Systems

Quantum computing has advanced rapidly to play a prominent role in potentially solving complex problems that have been beyond the ability of classical computing systems. Quantum computing systems of the future will increasingly be heterogeneous and distributed, spanning quantum-enabled, Post-Moore models and architectures, integrated classical and quantum processors as well as NISQ (Noisy Intermediate Scale Quantum) and post NISQ architectures. Such systems will require new, innovative, and visionary approaches to quantum computing hardware, quantum communication, heterogenous traditional and quantum system software and middleware, and consideration of human-computer interactions that are limited only by the imagination. These systems will need to be resilient to high qubit error rates and low decoherence times, adaptive to changing environments and usage scenarios, and able to meet the performance and energy objectives of applications such as quantum AI and machine learning, biotechnology, quantum and post-quantum cryptography, financial modeling and optimization, and many more. Computer systems support bridging the gap between applications and hardware in the form of operating system and runtime systems to manage and coordinate quantum hardware resources (e.g., qubit entanglement and magic states), execute error correcting kernels, manage quantum memory, and facilitate the execution of quantum algorithms is critical in realizing this potential. 

Embedded and Resource-Constrained Systems

Tremendous advancements have been made in IoT (Internet of Things) and edge computing technologies over the past decade. The challenge ahead is to develop software and sensing systems of the future, with humans in the loop, encompassing distributed, embedded, and resource-constrained devices that will span biotechnology, intelligent wearable computing, the Internet of Things (IoT), Augmented Reality/Virtual Reality (AR/VR), smart environments, intelligent transportation systems, personalized healthcare, data intensive and distributed systems, energy-efficient and sustainable computing, emerging workloads, digital twins, etc. These systems will need to be reliable and trustworthy, incorporate cutting-edge AI techniques and services, maintain privacy and security, adaptive to dynamically varying computing and communication environments as well usage scenarios, capable of supporting high-throughput applications, and able to meet performance and energy objectives. Clean-slate approaches considering security, privacy and energy-efficiency by design are particularly welcome.

Other Compute Architectures

Although digital computers typically employ the von Neumann computing paradigm, other approaches could be promising for next generation computing systems, including asynchronous or analog systems as well as neuromorphic, reconfigurable, dataflow, or other structures offering significant improvements. The incorporation of these approaches into the holistic computer systems ecosystem including processing, memory hierarchy and storage, interconnections and networking components, as well as with the software stack, presents research challenges and opportunities. Similarly, as new memory technologies mature, including volatile and non-volatile memories using semiconductors, neuromorphic, photonic, DNA, or quantum approaches, their effective integration into the computer systems ecosystem raises important research questions. Research proposals that address the impact of such technologies to computer systems are of interest, particularly creative, clean-slate approaches. 

 Areas of Research Interest

The CSR program seeks innovative research proposals that will advance

• the reliability, scalability, performance, power and energy-efficiency, security and privacy, scalability, and sustainability of quantum and classical computer systems;
• the infrastructure, software, and tools necessary to develop, manage, and execute algorithms on quantum or classical hardware, including operating systems, runtime systems, programming languages, simulators, and hardware interfaces specifically designed for quantum or classical computing;
• the infrastructure, software, and tools necessary to develop, manage, and execute the next generation of end-to-end AI systems and services as well as biotech applications, including holistic system designs encompassing sensing, learning, problem solving, perception, and prediction, as well as specialized hardware and software handling large datasets, automating workflows, and meeting regulatory compliance;
• seamless, transparent, and efficient parallel and distributed computation across classical and quantum computing systems, including scaling up from small embedded systems to supercomputers as well as scaling out from single computational nodes to highly parallel and distributed systems of many computational elements; and
• end-to-end storage and computing platforms encompassing edge and/or cloud to support the next generation of intelligent IoT applications and services, incorporating distributed, multi-modal sensing, real or near real-time processing, cutting-edge AI/intelligence technologies, data analytics and decision making, and security and privacy that will lead to innovative findings across research areas such as biotechnology, smart manufacturing, smart homes and transportation, and virtual/augmented reality.

In general, a research project focusing on computer system design, development, evaluation and/or deployment will be within the scope of the CSR program. However, research projects that focus on specific point solutions or specific application design and development or generic application development that does not leverage or advance systems architecture or the software stack may not be within its scope. Some examples of projects that are out of scope for the CSR program are projects focusing on: strengthening the security, privacy or usability of an existing computer system or system architecture; developing novel machine learning algorithms for a specific workload that will run on computer systems; applying machine learning or other scientific methods to solve a problem using computer systems as an evaluation platform; developing new algorithms that leverage the unique properties of quantum computers to solve a problem efficiently; or sensing technologies and cyber-physical systems with a closed control loop (making decisions based on sensing information, without human/computer systems in the loop). 

Note: Some projects might be appropriate for joint consideration by CSR and other programs across CISE and may be submitted for joint review.  It is advisable to put the additional programs in the title (for example, CSR: NeTS: ...).