Tag cloudcompilers sending security optimization Architecture environment aircraft hardware self-organization coverage faa sdr atc wireless prototype uas bia cyberphysical networking Synchronization simulation OFDM mobile deer location antenna FPGA bit-torrent program analysis directional privacy measurement anonymity modeling sampling phased array geostatistics optiscope cognitive radio Radio chainsaw pin
Tag Archives: Architecture
PI’s: Dirk Grunwald and Jeremy Siek
Students: Joseph Blomsted
Many complex applications contain mixed parallel behavior that can be expressed as a combination of task, data, and pipeline parallelism. To achieve the best performance, these applications must be carefully mapped to each target platform; the capabilities of the target system greatly influence the ratio of mixed-parallel behavior that should be exploited. Heterogeneous systems also require additional decisions concerning how code is scheduled across the available devices, as well as balancing the interdependence between device scheduling and parallelization strategy selection.
Conceptually, the program decomposition process occurs in three high-level phases. This holds true whether the decomposition is performed manually by a programmer or automatically through a parallel computing framework or compiler. First, code regions must be identified that have the appropriate granularity and execution characteristics to benefit from parallel execution. Then, a parallelization strategy (none, task, data, or pipeline) must be selected for each code region or group of related regions. Finally, the code regions must be converted into parallel code and scheduled to the available resources, either as static threads or entities within a parallel runtime system. This is often a challenging task because the appropriate parallelization strategy, granularity, and degree of nested parallelism for a region of code depends on the capabilities of the target system. The challenge is increased for heterogeneous parallel systems, which augment general-purpose computers with GPUs and other accelerators. These systems necessitate the additional task of scheduling regions across the different devices, and further complicate the parallelization strategy and granularity decisions
that are influenced by this schedule.
The Mixed-Parallel Heterogeneous Partitioning (MPHP) project is building a new program decomposition and scheduling algorithm that partitions a program into appropriately sized regions that are scheduled across the available resources using a mix of the three parallelization strategies. Unlike existing techniques, the presented algorithm does not evaluate the different parallelization strategies in a fixed order, but treats all three strategies equally. Furthermore, the system addresses the interdependence between device scheduling and strategy selection, reducing both decisions to a common problem that allows both issues to be explored simultaneously.
Wireless Radio Architecture
As we move forward towards the next generation of wireless protocols, the push for a better radio physical layer is ever increasing. Conventional radio architectures are limited to narrow operating regions and fails to adapt with changing technology. This is further strengthened with the advent of cognitive radio, which needs a more versatile and flexible framework that is programmable within the timing constraints of a protocol. In this research, we present an architecture for Software Defined Cognitive Radio that caters to the specific baseband processing requirements in a changing environment. We aim to provide more flexibility by deconstructing the radio pipeline into a framework of user controlled kernels that can be reconfigured at run-time. This architecture provides the barebones of a OFDM based radio physical layer that can adapt to perform a varied number of tasks in different radio networks. We also present a novel message based real-time reconfiguration method to transmit and receive a wide range of waveforms used in concurrent wireless protocols.
- Aveek Dutta, Dola Saha, Dirk Grunwald, Douglas Sicker, An Architecture for Software Defined Cognitive Radio, ACM/IEEE ANCS, La Jolla, USA, October 2010
- Aveek Dutta, Jeffery Fifield, Graham Schelle, Dirk Grunwald, Douglas Sicker, An Intelligent Physical Layer for Congnitive Radio Networks, ACM WICON, Maui, Hawaii, November 2008
- Jeff Fifield, Paul Kasemir, Dirk Grunwald and Douglas Sicker, Experiences With a Platform for Frequency-Agile Techniques, IEEE DySPAN 2007, Dublin, Ireland, April 2007.
Cyberphysical systems focus on the computer control control of physical systems. At Colorado, this has involved systems to control the energy or thermal characteristics of computer systems themselves as wel as the control of non-computer systems.
- Occam – A System for Adaptable Multicore Applications
- S-Taliro: Monte-Carlo Techniques for Testing Control Systems
- Infusion Pump Analysis Project
- Probabilistic Program Analysis Project.
- Relational Abstractions for Cyber-Physical Systems
- Flow*: Taylor Model Flowpipe Construction for Non-Linear Hybrid Systems
- Symbolic verification of hybrid systems
Selected Recent Publications
- Xin Chen, Erika Abraham and Sriram Sankaranarayanan. Flow*: An Analyzer for Non-Linear Hybrid Systems , Computer-Aided Verification (CAV) 2013
- Aleksandar Chakarov and Sriram Sankaranarayanan. Program Analysis with Martingales , Computer-Aided Verification (CAV) 2013
- Sriram Sankaranarayanan, Aleksandar Chakarov and Sumit Gulwani. Static Analysis of Probabilistic Programs: Inferring Whole Program Properties from Finitely Many Executions, ACM Conference on Programming Language Design and Implementation (PLDI), 2013
- Paul Givens, Aleksandar Chakarov, Sriram Sankaranarayanan and Tom Yeh. Exploring the Internal State of User Interfaces by Combining Computer Vision Techniques with Grammatical Inference. Intl. Conference on Software Engg. (ICSE), New Ideas and Emerging Research Track, 2013
- Sriram Sankaranarayanan, Chris Miller, Rangarajan Ragunathan, Hadi Ravanbakhsh and Georgios Fainekos. A Model-Based Approach to Synthesizing Insulin Infusion Pump Usage Parameters for Diabetic Patients. Allerton 2012 conference proceedings , 2012 (invited paper ).
- Xin Chen, Erika Abraham and Sriram Sankaranarayanan. Taylor Model Flowpipe Construction for Non-linear Hybrid Systems. IEEE Real-Time Systems Symposium (RTSS), 2012 (Supplementary Materials: here).
- Sriram Sankaranarayanan and Georgios Fainekos. Simulating Insulin Infusion Pump Risks by In-Silico Modeling of the Insulin-Glucose Regulatory System. Computational Methods in Systems Biology (CMSB’12) 2012.
- Aditya Zutshi, Sriram Sankaranarayanan and Ashish Tiwari. Timed Relational Abstractions of Sampled-Data Control Systems , Computer-Aided Verification (CAV’12), 2012 (to appear). Supplementary Materials are available here .
- Georgios Fainekos, Sriram Sankaranarayanan, Koichi Ueda and Hakan Yazarel. Verification of Automotive Control Systems using S-Taliro , Proc. of American Control Conference (ACC’12) invited session on verification of automotive control systems.
- Arlen Cox, Sriram Sankaranarayanan and Bor-Yuh Evan Chang. A Bit Too Precise? Bounded Verification of Quantized Digital Filters. Tools and Algorithms for Construction and Analysis of Systems (TACAS) 2012.
- Aleksandar Chakarov, Sriram Sankaranarayanan and Georgios Fainekos. Combining Time and Frequency Domain Specifications For Periodic Signals, Runtime Verification (RV) 2011.
- Sriram Sankaranarayanan and Ashish Tiwari, Relational Abstraction for Continuous and Hybrid Systems, Computer-Aided Verification (CAV), 2011. Supplementary materials are available here .
- D. Fay, L. Shang, and D. Grunwald, “A platform for developing adaptable multicore applications,” in Proc. IEEE International Conference on Compilers, Architecture, and Synthesis for Embedded Systems, July 2009.
- Michael Colon and Sriram Sankaranarayanan, Generalizing the Template Polyhedral Domain, European Symp. on Programming (ESOP’11), 2011.
- Sriram Sankaranarayanan, Automatic Abstraction of Non-Linear Systems Using Change of Variables Transformations, Hybrid Systems: Computation and Control (HSCC’11), 2011.