Search by title, author, keywords, etc.
We have developed a learning environment where students teach a computer agent, using visual representations, and can monitor the agent’s learning progress by asking her questions and having her take quizzes. The system provides self-regulated learning and metacognitive support via dialog-embedded prompts from Betty, the teachable agent, and Mr. Davis, the mentor agent. Our primary goals have been to support learning of complex science topics in middle school classrooms and facilitate development of metacognitive skills to support future learning. In this paper, we discuss methods that we have employed for detecting and characterizing students’ behavior patterns from their activity sequences on the system. In particular, we discuss a method for learning hidden Markov models (HMM) from the activity logs. We demonstrate that the HMM structure corresponds to students’ aggregated behavior patterns in the learning environment. Overall, the HMM technique allows us to go beyond simple frequency and sequence analyses, such as individual activity and pre-defined pattern counts, instead using exploratory methods to examine how these activities cohere in larger patterns over time. The paper outlines a study conducted in a 5th grade science classroom, presents the models derived from the students’ activity sequences, interprets the model structure as aggregate patterns of their learning behaviors, and links these patterns to students’ use of self-regulated learning strategies. The results illustrate that those who teach an agent demonstrate better learning performance and better use of metacognitive monitoring behaviors than students who only learn for themselves. We also observed more advanced and focused monitoring behaviors in the students who received metacognitive strategy feedback from the mentor agent while they taught the teachable agent.
Authored by Gautam Biswas, Hogyeong Jeong, John Kinnebrew, Brian Sulcer, and Rod Roscoe
Authored by Constantin Aliferis, Alexander Statnikov, Ioannis Tsamardinos, Subramani Main, and Xenofon Koutsoukos
Authored by Constantin Aliferis, Alezander Statnikov, Ioannis Tsamardinos, Subramani Main, and Xenofon Koutsoukos
Authored by Gabor Karsai, Gregor Engels, Claus Lewerentz, Wilhelm Schäfer, Andy Schürr, and Bernhard Westfechtel
Authored by Rajat Mehrotra, Abhishek Dubey, Sherif Abdelwahed, and Asser Tantawi
Authored by Matthew Daigle, Xenofon Koutsoukos, and Gautam Biswas
Large scientific computing data-centers require a distributed dependability subsystem that can provide fault isolation and recovery and is capable of learning and predicting failures to improve the reliability of scientific workflows. This paper extends our previous work on the scientific workflow management systems by presenting a hierarchical dynamic workflow management system that tracks the state of job execution using timed state machines. Workflow monitoring is achieved using a reliable distributed monitoring framework, which employs publish-subscribe middleware built upon OMG Data Distribution Service Standard. Failure recovery is achieved by stopping and restarting the failed portions of workflow directed acyclic graph.
Authored by Pan Pan, Abhishek Dubey, and Luciano Piccoli
Authored by Jia Bai, Emeka Eyisi, Yuan Xue, and Xenofon Koutsoukos
Timed failure propagation graph (TFPG) is a directed graph model that represents temporal progression of failure effects in physical systems. In this paper, a distributed diagnosis approach for complex systems is introduced based on the TFPG model settings. In this approach, the system is partitioned into a set of local subsystems each represented by a subgraph of the global system TFPG model. Information flow between subsystems is achieved through special input and output nodes. A high level diagnoser integrates the diagnosis results of the local subsystems using an abstract high level model to obtain a globally consistent diagnosis of the system.
Authored by Nagabhushan Mahadevan, Sherif Abdelwahed, Abhishek Dubey, and Gabor Karsai
Authored by Aniruddha Gokhale, Mark McDonald, Steven Drager, and William McKeever
Authored by Matthew Daigle, Indranil Roychoudhury, Gautam Biswas, Xenofon Koutsoukos, Ann Patterson-Hine, and Scott Poll
Authored by Jonathan Wellons, Liang Dai, Yuan Xue, and Yi Cui
Authored by J. Sagedy, B. Sulcer, and G. Biswas
Authored by Fan Qiu and Yi Cui
Authored by Bo Liu, Yanchuan Cao, Yi Cui, Yansheng Lu, and Yuan Xue
Authored by Laszlo Juracz, Gabor Pap, Larry Howard, and Julie Johnson
Planning and scheduling for agents operating in heterogeneous, multi-agent environments is governed by the nature of the environment and the interactions between agents. Significant efficiency and capability gains can be attained by employing planning and scheduling mechanisms that are tailored to particular agent roles. This paper presents such a framework for a global sensor web that operates as a two-level hierarchy, where the mission level coordinates complex tasks globally and the resource level coordinates the operation of subtasks on individual sensor networks. We describe important challenges in coordinating among agents employing two different planning and scheduling methods and develop a coordination solution for this framework. Experimental results validate the benefits of employing guided, context-sensitive coordination of planning and scheduling in such sensor web systems.
Authored by John Kinnebrew, Daniel Mack, Gautam Biswas, and Douglas Schmidt
This paper recalls a non-linear constructive method, based on controlling cascades of conic-systems as it applies to the control of quad-rotor aircraft. Such a method relied on the physical model of the system to construct high performance, modest sampling period (Ts = .02 s) and low complexity digital-controllers. The control of fixed-wing aircraft, however is not nearly a straight forward task in extending results related to the control of quad-rotor aircraft. Although fixed-wing aircraft and quad-rotor aircraft ultimately share the same kinematic equations of motion, fixed-wing aircraft are intimately dependent on their relationship to the wind reference frame. This additional coupling leads to additional equations of motion including those related to the angle-of attack, slide-slip-angle, and bank angle. As a result a more advanced non-linear control method known as back-stepping is required to compensate for non-passive non-linearity's. These back-stepping controllers are recursive in nature and can even address actuator magnitude and rate limitations and even include adaptability to unknown lift and drag coefficients. This paper presents a non-adaptive back-stepping controller which is aimed to verify a fixed-wing aircraft model not subject to actuator limitations (in order to simplify discussion). The back-stepping controller proposed is less complex then previously proposed controllers, exhibits similar response characteristics while being robust to both steady head wind shear and discrete-time wind gust disturbances.
Authored by Nicholas Kottenstette
Several localization algorithms exist for wireless sensor networks that use angle of arrival measurements to estimate node position. However, there are limited options for actually obtaining the angle of arrival using resource-constrained devices. In this paper, we describe a radio interferometric technique for determining bearings from an anchor node to any number of target nodes at unknown positions. The underlying idea is to group three of the four nodes that participate in a typical radio interferometric measurement together to form an antenna array. Two of the nodes transmit pure sinusoids at close frequencies that interfere to generate a low-frequency beat signal. The phase difference of the measured signal between the third array node and the target node constrains the position of the latter to a hyperbola. The bearing of the node can be estimated by the asymptote of the hyperbola. The bearing estimation is carried out by the node itself, hence the method is distributed, scalable and fast. Furthermore, this technique does not require modification of the mote hardware because it relies only on the radio. Experimental results demonstrate that our approach can estimate node bearings with an accuracy of approximately 3 degrees in 0.5 sec.
Authored by Isaac Amundson, Janos Sallai, Xenofon Koutsoukos, and Akos Ledeczi
This paper proposes a combined energy-based and physics of failure model for degradation analysis and prognosis of electrolytic capacitors in DC-DC power converters. Electrolytic capacitors and MOSFET’s have higher failure rates than other components in DC-DC converter systems. Currently our work focuses on analyzing and modeling electrolytic capacitors degradation and its effects on the output of DC-DC converter systems. The output degradation is typically measured by the increase in ripple current and the drop in output voltage at the load. Typically the ripple current effects dominate, and they can have adverse effects on downstream components. For example, in avionics systems where the power supply drives a GPS unit, ripple currents can cause glitches in the GPS position and velocity output, and this may cause errors in the Inertial Navigation (INAV) system causing the aircraft to fly off course. A model based approach to studying degradation phenomena enables us to combine the energy based modeling of the DC-DC converter with physics of failures models of capacitor degradation, and predict using stochastic simulation methods how system performance deteriorates with time. This more systematic analysis may provide a more general and accurate method for computing the remaining useful life (RUL) of the component and the converter system. We have employed a topological energy based modeling scheme based on the bond graph (BG) modeling language for building parametric models of multi-domain physical systems. The BG approach captures relationship between component parameters, system behavior and performance. Component degradation models are constructed using empirical physics of failure models that have been presented in the literature, and validating these models using data collected from accelerated degradation studies. The physics of failure models provide mathematical formulations that are directly linked to component parameters. Literature reports a number of operating conditions that may cause capacitor degradation. These include High Voltage conditions, Transients, Reverse Bias, Strong Vibrations and high ripple current. In our work, we have studied the effects of capacitor degradation on DC-DC converter performance by developing a combination of converter system model and a physics of failure model of electrolytic capacitor degradation when subjected to thermal and electrical stresses. Thermal stress occurs when the capacitors operate in high temperature environments, while electrical stress conditions occur due to high operating voltages and even ripple currents above the rated values. In our work we are developing models to capture the failure phenomenon in these components. Our current work adopts a physics of failure model (Arrhenius Law) for equivalent series resistance (ESR) increase in electrolytic capacitors subjected to electrical and thermal stresses. Under stress conditions the ESR gradually increases and capacitance of the capacitor gradually decreases with time thus resulting in the capacitors ability to filter out AC components in the output voltage. As a result, the output ripples current and ripple voltages of the converter increases over time. The output DC voltage also decreases over time, but the ripple current effects on the load are more significant. High ripple currents may lead to frequent resets or even damage in the systems that are downstream from the power supply. We present a combined model- and data-driven approach for estimating and validating the parameters of our physics of failure models for capacitor degradation. We use Monte Carlo simulation methods to develop prognostic methods that predict remaining useful life based on degradation in the power supply output. For model simulation study the derived degradation model of the capacitors are reintroduced into the DC-DC converter system model to study changes in the system performance using Monte Carlo methods. The simulation results observed under different stress conditions are recorded and compared with the hardware experiments. We have designed different hardware setups for capturing the data of actual degradation phenomenon under thermal and electrical stress. In the first setup capacitors are subjected to only thermal stress. Under this condition output ripple voltage and increase in ESR is monitored over the period of time. In the second setup experiment the capacitors are subjected to electrical stress by continuous charging/discharging cycle. The ESR parameter is monitored regularly over this period of time. The data from these experiments is used to verify results from the models developed and also for refining the model parameters for more accuracy. The paper concludes with comments and future work to be done.
Authored by Chetan Kulkarni, Gautam Biswas, Xenofon Koutsoukos, Goebel Kai, and Celaya Jose
Current methods of estimating air quality involve assigning a single value called the Air Quality Index (AQI) to a large land area for a 24-hour period based on a very few, sparsely-located sensors. This produces a low-resolution image of the air quality in that region. We have devised a new mobile air quality monitoring network with the ability to provide high-resolution realtime pollution data at any location within the coverage area. We have prototyped sensors and proven the feasibility of the approach, and are currently testing a small-scale implementation using irregularly-sampled spatiotemporal measurements from mobile car-mounted sensors coupled with existing static data. This data feeds a web-based application, enabling users to view air quality in specific regions as well as estimate exposure over speci ed time periods and plan routes based on minimal exposure to a given pollutant.
Authored by Will Hedgecock, Peter Volgyesi, Akos Ledeczi, Xenofon Koutsoukos, and Akram Aldroubi
Authored by Sandor Szilvasi, Janos Sallai, Isaac Amundson, Peter Volgyesi, and Akos Ledeczi
Network security is a major issue affecting SCADA systems designed and deployed in the last decade. Simulation of network attacks on a SCADA system presents certain challenges, since even a simple SCADA system is composed of models in several domains and simulation environments. Here we demon- strate the use of C2WindTunnel to simulate a plant and its controller, and the Ethernet network that connects them, in different simulation environments. We also simulate DDOS-like attacks on a few of the routers to observe and analyze the effects of a network attack on such a system.
Authored by Rohan Chabukswar, Bruno Sinopoli, Gabor Karsai, Annarita Giani, Himanshu Neema, and Andrew Davis
Authored by Ethan Jackson, Wolfram Schulte, Daniel Balasubramanian, and Gabor Karsai