Computer Engineering
Adaptive Hardware and Systems on Reconfigurable Chips
A complete useful adaptive system on FPGA, contains an embedded processor cores along with all other functional hardware components such as memories, input/output and communication interfaces. Adaptation reflects the capability of a system to maintain or improve its performance in the context of internal or external changes such as environmental changes, interference, modification of requirements, trade-offs between performance and resources. In this research, we investigate the use of partial reconfiguration capabilities of FPGAs, and adaptive and evolvable hardware design techniques and platforms in the design and implementation of such systems. These systems are applied to several real world solutions such as: health diagnosis, networks, recognition, identification, inspection, automation, networks, and control.
Neuromorphic Architectures
Neuromorphic architectures have the potential to revolutionize computing efficiency and capability. Drawing inspiration from biology, neuromorphic systems can learn, adapt, and process diverse sets of information originating from a variety of sources. The overall goal is the design of custom hardware solutions to implement various machine learning algorithms. These systems may rely on analog signals, voltage pulses/spikes, or other physical phenomena for information transmission.
MORE INFO
To learn more, visit the Cantley Research Group online.
PARTICIPATING ECE FACULTY
- Kurtis Cantley
FACILITIES
The Electronic and Neuromorphic Device and Systems (ENDS) Lab is located in the Micron Engineering Center, room 312. The Transport Characterization Lab and the Idaho Microfabrication Laboratory are also utlized in this research.
FUNDING
- Defense Threat Reduction Agency contract HDTRA11710036: “Impact of Radiation on Spatio-Temporal Pattern Recognition in Memristor-Based Neuromorphic Circuits”
- National Science Foundation award 1751230: “CAREER: Spiking Neural Circuits and Networks with Temporally Dynamic Learning”
COLLABORATION CAPACITY
This research has opportunities for collaboration in multiple areas of computer science, specifically software machine learning, firmware development, and operating systems
RELEVANT PUBLICATIONS
- R. Ivans and K. D. Cantley, “A Spatiotemporal Pattern Detector,” in 2019 IEEE 62nd International Midwest Symposium on Circuits and Systems (MWSCAS), 2019, pp. 444–447.
- S. G. Dahl, R. C. Ivans, and K. D. Cantley, “Learning Behavior of Memristor-Based Neuromorphic Circuits in the Presence of Radiation,” in Proceedings of the International Conference on Neuromorphic Systems, 2019.
- S. G. Dahl, R. C. Ivans, and K. D. Cantley, “Radiation Effect on Learning Behavior in Memristor-Based Neuromorphic Circuit,” in 2019 IEEE 62nd International Midwest Symposium on Circuits and Systems (MWSCAS), 2019, pp. 53–56.
- S. G. Dahl, R. Ivans, and K. D. Cantley, “Modeling Memristor Radiation Interaction Events and the Effect on Neuromorphic Learning Circuits,” Proc. Int. Conf. Neuromorphic Syst. – ICONS ’18, pp. 1–8, 2018.
- K. D. Cantley, R. C. Ivans, A. Subramaniam, and E. M. Vogel, “Spatio-Temporal Pattern Recognition in Neural Circuits with Memory-Transistor-Driven Memristive Synapses,” in 2017 International Joint Conference on Neural Networks (IJCNN), 2017, pp. 4633–4640.
Sensor Systems
Sensor nodes are composed of small microprocessors connected to a set of sensors and supported by various communications, power management, and data storage systems. Sensor nodes must often be power efficient as they may need to operate for extended periods without external power. The microprocessors used in embedded systems are inherently limited in comparison to the processors inside desktop computers. Embedded processors give up some overall processing power in exchange for both reduced physical size and reduced power requirements. Current research include FAA airliner cabin environmental monitoring and NIH in-home air quality monitoring.
Spiking Neural Networks
This research involves the use of novel semiconductor materials and devices such as memristors for implementing electronic neural networks with biologically realistic behaviors and learning modalities. The motivation for this effort is that the brain is able to process many types of real-world information far more efficiently than today’s digital computers. It is also adept at experiential learning and making predictions and inferences, which are critical properties for future intelligent computers.
MORE INFO
To learn more, contact Kris Campbell (kriscampbell@boisestate.edu) for device research, and Kurtis Cantley (kurtiscantley@boisestate.edu) for circuit research, or visit https://www.boisestate.edu/coen-crg
PARTICIPATING ECE FACULTY
- Kris Campbell
- Kurtis Cantley
FACILITIES
Device fabrication is performed in the Idaho Microfabrication Laboratory. Electrical characterization is performed using state-of-the-art microprobe stations in various labs. More information on electrical characterization can be found on Kris Campbell’s research site.
COLLABORATION CAPACITY
Theoretical neuroscience and cognitive sciences are two potential fields with strong research collaboration potential
FUNDING
- National Science Foundation award 1751230: “CAREER: Spiking Neural Circuits and Networks with Temporally Dynamic Learning”
RELEVANT PUBLICATIONS
- Campbell, K. A.; Drake, K. T.; and Barney Smith, E. H. “Pulse shape and timing dependence on the spike-timing-dependent plasticity response of ion-conducting memristors as synapses,” Frontiers in Bioengineering and Biotechnology (2016) 4, article 97, 1-11. Open access: http://journal.frontiersin.org/article/10.3389/fbioe.2016.00097/full
- R. C. Ivans, S. Member, S. G. Dahl, S. Member, and K. D. Cantley, “A Model for R(t) Elements and R(t)-Based Spike-Timing-Dependent Plasticity With Basic Circuit Examples,” IEEE Trans. Neural Networks Learn. Syst., vol. PP, pp. 1–11, 2019.
- R. C. Ivans, K. D. Cantley, and J. L. Shumaker, “A CMOS Synapse Design Implementing Tunable Asymmetric Spike Timing-Dependent Plasticity,” in International Midwest Symposium on Circuits and Systems (MWSCAS), 2017, pp. 0–3.
Wireless Sensor Networks
Frequently, individual sensor nodes are connected wirelessly to form wireless sensor networks. Networking sensor nodes allows scientists to more accurately characterize phenomenon of interest by providing a means to both temporally and spatially relate sensor data.