An official website of the United States government
Internet-of-things (IoT) introduce new attack surfaces for power grids with the usage of Wi-Fi enabled high wattage appliances. Adversaries can use IoT networks as a foothold to significantly change load demands and cause physical disruptions in power systems. This new IoT-based attack makes current security mechanisms, focusing on either power systems or IoT clouds, ineffective. To defend the attack, we propose to use a data-centric edge computing infrastructure to host defense mechanisms in IoT clouds by integrating physical states in decentralized regions of a power grid. By enforcing security policies on IoT devices, we can significantly limit the range of malicious activities, reducing the impact of IoT-based attacks. To fully understand the impact of data-centric edge computing on IoT clouds and power systems, we developed a cyber-physical testbed simulating six different power grids. Our preliminary results show that performance overhead is negligible, with less than 5% on average.
more »
« less
Poster: Identity-Independent IoT for Overarching Policy Enforcement
Enforcing overarching policies such as safety norms and energy restrictions becomes critical as IoT scales and integrates into large systems. These policies should be applied preemptively and capable of adapting to system changes. Traditional IoT systems, reliant on fixed device identities, limit reliability, scalability, and resilience. Thus, we propose Identity-Independent IoT (I3oT), centered on adopting flexible descriptors to enforce policies. I3oT introduces a separate management plane on top of the standard operational workflow, thereby enhancing safety in scalable and integrated IoT systems.
more »
« less
- Award ID(s):
- 1932418
- PAR ID:
- 10525942
- Publisher / Repository:
- IEEE
- Date Published:
- ISBN:
- 979-8-3503-5487-4
- Page Range / eLocation ID:
- 296 to 296
- Format(s):
- Medium: X
- Location:
- San Francisco, CA, USA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
The increasing complexity and ubiquity of using IoT devices exacerbate the existing programming challenges in smart environments such as smart homes, smart buildings, and smart cities. Recent works have focused on detecting conflicts for the safety and utility of IoT applications, but they usually do not emphasize any means for conflict resolution other than just reporting the conflict to the application user and blocking the conflicting behavior. We propose RemedIoT, a remedial action 1 framework for resolving Internet-of-Things conflicts. The RemedIoT framework uses state of the art techniques to detect if a conflict exists in a given set of distributed IoT applications with respect to a set of policies, i.e., rules that define the allowable and restricted state-space transitions of devices. For each identified conflict, RemedIoT will suggest a set of remedial actions to the user by leveraging RemedIoT's programming abstractions. These programming abstractions enable different realizations of an IoT module while safely providing the same level of utility, e.g., if an air-conditioner application that is used to implement a cooling module conflicts with a CO2 monitor application that requires ventilation at home, a non-conflicting smart fan application will be suggested to the user. We evaluate RemedIoT on Samsung SmartThings applications and IFTTT applets and show that for 102 detected conflicts across 74 sample applications with 11 policies, RemedIoT is able to remediate ~ 80% of the conflicts found in the environment, which would normally be blocked by prior solutions. We further demonstrate the efficacy and scalability of our approach for smart city environments.more » « less
-
The objective of this research is to enable safety‐critical systems to simultaneously learn and execute optimal control policies in a safe manner to achieve complex autonomy. Learning optimal policies via trial and error, that is, traditional reinforcement learning, is difficult to implement in safety‐critical systems, particularly when task restarts are unavailable. Safe model‐based reinforcement learning techniques based on a barrier transformation have recently been developed to address this problem. However, these methods rely on full‐state feedback, limiting their usability in a real‐world environment. In this work, an output‐feedback safe model‐based reinforcement learning technique based on a novel barrier‐aware dynamic state estimator has been designed to address this issue. The developed approach facilitates simultaneous learning and execution of safe control policies for safety‐critical linear systems. Simulation results indicate that barrier transformation is an effective approach to achieve online reinforcement learning in safety‐critical systems using output feedback.more » « less
-
Privacy policies contain important information regarding the collection and use of user’s data. As Internet of Things (IoT) devices have become popular during the last years, these policies have become important to protect IoT users from unwanted use of private data collected through them. However, IoT policies tend to be long thus discouraging users to read them. In this paper, we seek to create an automated and annotated corpus for IoT privacy policies through the use of natural language processing techniques. Our method extracts the purpose from privacy policies and allows users to quickly find the important information relevant to their data collection/use.more » « less
-
Modern Internet of Things (IoT) applications, from contextual sensing to voice assistants, rely on ML-based training and serving systems using pre-trained models to render predictions. However, real-world IoT environments are diverse, with rich IoT sensors and need ML models to be personalized for each setting using relatively less training data. Most existing general-purpose ML systems are optimized for specific and dedicated hardware resources and do not adapt to changing resources and different IoT application requirements. To address this gap, we propose MLIoT, an end-to-end Machine Learning System tailored towards supporting the entire lifecycle of IoT applications. MLIoT adapts to different IoT data sources, IoT tasks, and compute resources by automatically training, optimizing, and serving models based on expressive applicationspecific policies. MLIoT also adapts to changes in IoT environments or compute resources by enabling re-training, and updating models served on the fly while maintaining accuracy and performance. Our evaluation across a set of benchmarks show that MLIoT can handle multiple IoT tasks, each with individual requirements, in a scalable manner while maintaining high accuracy and performance. We compare MLIoT with two state-of-the-art hand-tuned systems and a commercial ML system showing that MLIoT improves accuracy from 50% - 75% while reducing or maintaining latency.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1109/SPW63631.2024.00036
