Intermittent Computing On Batteryless IoT Devices
The shift from a cloud-centric to a thing/data-centric approach, embodied by the Internet of Things (IoT), has the potential to solve issues like latency, scalability, privacy, and security. The IoT market is projected to reach $4.5 trillion by 2035, covering industries from smart homes to environmental monitoring. Despite its growth, battery-powered IoT devices face limitations such as short lifespan, high cost, and environmental impact, with global battery disposal expected to hit 78 million units daily by 2025.
Innovations like batteryless devices using energy-harvested systems and Edge AI, endorsed by organizations like the NSF and Department of Energy, can address these issues. Effective solutions will require rethinking conventional architectures, emphasizing Intelligent Elastic Intermittent Computation (EIC) devices. These devices promise to overcome the limitations of current IoT systems through advancements in energy-harvested, normally-off computing and intermittent-aware machine learning algorithms. The research goal is to develop duty-cycle-variable computing to create sustainable and efficient IoT systems.
Security and Robustness for Intermittent Computing Using Cross-Layer Post-CMOS Approaches
Energy-harvesting, resource-limited devices used in IoT applications, such as wearables and industrial systems, are vulnerable to various attacks, including side-channel and magnetic attacks, as well as power outages. These vulnerabilities could cause serious harm due to the vast number of connected devices. Current defense mechanisms either have high overheads, making them unsuitable for resource-constrained devices, or are incomplete.
This research focuses on developing secure, lightweight strategies for normally-off, energy-harvesting IoT nodes. By utilizing emerging non-volatile, spin-based devices, the goal is to construct reconfigurable, secure logic circuits that maintain robust computation even during attacks. These circuits will store intermediate states within non-volatile devices to ensure continuous processing. The research aims to integrate circuit design, architecture, and algorithmic techniques to bolster security while minimizing resource overhead, creating a more resilient and efficient IoT environment. This work sets the foundation for a safer, more secure future in IoT applications.
Cross-Layer Solutions Enabling Instant Computing for Edge Intelligence Devices
This project focuses on developing low-cost, efficient design strategies for real-time processing and decision-making systems across multiple design abstraction levels. Key objectives include:
-
Designing low-power, area-efficient integrated converters, such as Binary-to-Residue Number Systems (RNS) and Analog-to-Digital Converters (ADC).
-
Proposing a reconfigurable near-sensor RNS processing unit (RPU) to accelerate low-bit-width neural networks, enabling intelligent IoT devices. The RPU can perform MAC operations with one-cycle latency while maintaining memory function and capacity, handling essential layers like activation and normalization.
-
Creating an automated exploration tool called Residual Architecture Search (RAS) to optimize processing units' architecture for various applications, improving metrics like energy efficiency and performance.
-
Developing a cross-layer evaluation framework, including FPGA-magnetic RAM prototypes, and testing architectures with RISC-V processors on real-world IoT applications.
This project aims to enhance IoT device performance while reducing energy consumption and boosting resilience through innovative design techniques.
Integrated Sensing and Normally-off Computing for Edge Imaging Systems
The Internet of Things (IoT) market is expected to reach $1100B by 2025, with over 75 billion interconnected devices. These devices collect vast amounts of sensory data, much of which is redundant and unstructured. The process of data conversion, storage, and computation in IoT devices leads to significant energy consumption, latency, and memory bottlenecks. Additionally, the cost and impracticality of replacing batteries in IoT devices drive the need for energy-harvesting solutions.
This project addresses these challenges by developing computing architectures that are high-speed, energy-efficient, and normally-off under regular conditions. The research focuses on integrating sensing computation for resource-limited sensory nodes, combining cross-layer post-CMOS methodologies with non-volatile devices. By leveraging always-on sensing and normally-off computing, and ensuring intermittent-robust computation through secure storage of intermediate circuit values, this work aims to revolutionize IoT device operations, making them more resilient, energy-efficient, and capable of handling power outages.
