Program Schedule: PhD Forum 1-3 and New Investigators 1-2
Wednesday, October 1st
10:00 a.m. — 1:00 p.m.
PhD Forum 1 - Interdisciplinary Computing
Location: Crestone Peak II-IV
Towards Compact, Robust DNA Self-Assembly based Computation: Modeling, Simulation and Experiments
Presenter: Urmi Majumder (Duke University)
Although self-assembly is a powerful technique for constructing nanoscale objects, due to its great complexity, its immense engineering potential has been inadequately harnessed. My thesis addresses this challenge by asking the following question: How can self-assembly be used to perform arbitrarily complex and robust computation? I approach this question by studying basic mathematical properties of self-assembly and by designing, simulating and fabricating self-assembled systems using DNA as a nanoconstruction material.
Computational Modeling and Formal Analysis Techniques in Interdisciplinary Studies of Complex Systems
Presenter: Mona Vajihollahi (Simon Fraser University)
This work deals with the challenges of applying computational techniques in novel research areas, such as Computational Criminology, focusing on the needs of truly interdisciplinary research projects. Bridging the gap between disciplines requires new approaches that accommodate solid common grounds for clarifying different assumptions and expectations. We present a methodological framework that addresses such special needs, and focuses on the ’cooperative process’ of transforming complex domain knowledge into computational artifacts.
Spatial Analysis of Meeting Speech Scenes
Presenter: Eva Cheng (University of Wollongong)
Within the field of digital speech signal processing, this doctoral research focuses on speech signal description and extraction of semantically meaningful metadata. Of particular interest is the use of spatial information to detect and annotate significant events in multichannel meeting speech recordings. The research methodology adopted approaches the problem of reliably estimating speaker location information through developing software algorithms, investigating multi-microphone hardware solutions, and optimally combining the two together.
PhD Forum 2 - Hardware, Real-time & Embedded Systems
Location: Torreys Peak II
Analytically Bounding Data Cache Behavior for Real-Time Systems
Presenter: Harini Ramaprasad (North Carolina State University)
My research work makes contributions to data cache analysis and static timing analysis for hard real-time systems. First, data cache behavior is characterized statically for a single task. Second, data cache preemption effects are accounted for, thus calculating tight response time bounds for preemptive tasks. Third, a methodology is proposed to calculate an upper bound on the response times of tasks in systems where some tasks may have critical sections.
Leveraging Mixed-Process 3D Integration for Reliability and Cache Hierarchies
Presenter: Niti Madan (University of Utah)
Emerging three-dimensional (3D) integration technology enables vertical stacking of silicon dies with high density and low latency interconnects. This results in increased processor performance as well as reduced power consumption because of smaller on-chip wires. Several research opportunities and challenges in 3D technology are being studied at the fabrication, circuit design and architecture level. This dissertation explores novel applications for 3D die stacking at the micro-architecture level with an emphasis on mixed-process integration.
Embedded Systems in Body Sensor Networks: Experimental Methods for Continuous Monitoring, Network Connectivity, and Physical Security
Presenter: Tammara Massey (University of California, Los Angeles)
The shrewd design of embedded systems make efficient use of limited resources through efficient reconfiguration schemes that balance the trade-offs between power consumption, memory consumption, and interoperability in heterogeneous environments. Medical applications, in particular, will benefit from lightweight reconfiguration techniques that will improve the quantity and quality of ubiquitous data collection. Reconfigurable software in heterogeneous embedded systems that adapts to the environment is essential for usability and system performance.
PhD Forum 3 - Mixed Session: Sensors, Sensor Networks & User Interfaces
Location: Torreys Peak III
Context-Sensitive Intelligent Cueing
Presenter: Julie S. Weber (University of Michigan)
I focus on evaluating a variety of visual presentations of electronic notifications. In contrast to the breadth of work addressing the question of when to interrupt a computer user with a notification, my work addresses the less studied question of how to deliver interruptions. Results motivate future investigation into the efficacy of varying the presentation style of a notification for tailoring to users’ individual preferences or requirements.
Quality of Information-aware Design and Management of Sensor Network
Presenter: Sadaf Zahedi (University of California, Los Angeles)
Providing a high quality of information (QoI) from sensor networks is an important design goal. A modular analysis framework is introduced to evaluate the QoI of sensor network deployments. The process is decomposed to steps of modeling the characteristics of sensor networks, analyzing the QoI at sensor and network level, and exploring trade-offs and optimized designs. Sensor network is managed at runtime to detect the faults in tiered fashion and take run-time actions.
Multitier Multiscale Sensing: A New Paradigm for Actuated Sensing
Presenter: Diane M Budzik (University of California, Los Angeles)
Multitier Multiscale Sensing is a new paradigm for actuated sensing for efficiently sampling dynamic spatiotemporal phenomena with high Þdelity. This approach introduces a hierarchy of sensors according to sampling Þdelity, spatial coverage, and mobility characteristics. The application of solar light radiation illustrates a two-tier implementation of multiscale sensing. Experiments performed in simulation and on a physical robotic system show that multitier multiscale sensing is suitable for sampling dynamic spatiotemporal phenomena.
New Investigators 1 - Networking
Location: Torreys Peak IV
A 3D N-tier Architecture for a Multi-parameter Ocean Monitoring Underwater Wireless Sensor Network
Presenters: Supriya Vadlamani (Birla Institute of Technology and Science), Aparna Sasidharan (Birla Institute of Technology and Science)
In this paper we propose a 3D multi-tier architecture for Underwater Wireless Sensor Networks which can be used for Multi-parameter Ocean column monitoring. The nodes in this network can be fixed or mobile. The network topology can change with node mobility and the network is reconfigurable. The architecture defines multiple layers along the entire ocean column. A varying combination of sensor clusters and different underwater vehicles is employed at each level.
Minerva: Learning to Infer Network Path Properties
Presenter: Rita H Wouhaybi (Intel Corporation)
Knowledge of the network path properties such as latency, and hop count is key to the performance of overlay networks, grids and p2p applications. However, the size of the Internet makes the task of measuring these immensely difficult. In this paper, we propose a novel learning-based approach, called Minerva, for the inferencing of inter-node properties, resulting in a more scalable approach based on partial measurements.
New Investigators 2 - Modeling & Simulation
Location: Crestone Peak I
Efficient Parallel Simulation of an Individual-Based Fish Schooling Model on a Graphics Processing Unit
Presenters: Hong Li (University of California, Santa Barbara), Allison Kolpas (University of California, Santa Barbara)
Due to their low cost and high performance processing capabilities, graphics processing units (GPUs) have become an attractive alternative to clusters for some scientific computing applications. In this paper we show how stochastic simulation of an individual-based fish schooling model can be efficiently carried out on a general-purpose GPU. We describe our implementation and present computational results to illustrate the power of this new technology.
Minimum Cost Routing with Local Processing for Distributed Statistical Inference
Presenter: Animashree Anandkumar (Cornell University)
Classical routing does not exploit “inherent” saving in costs arising from data reduction in a sufficient statistic for inference. We explore in-network processing for inference, using Markov random field (MRF) model for spatial correlation. We show that the minimum cost routing for computation and delivery of the likelihood ratio is an approximation-ratio preserving Steiner tree on expanded graph. Hence, any Steiner-tree approximation has the same ratio for minimum cost fusion.
A Remote Server-based Network Emulation System
Presenter: Yan Gu (North Dakota State University)
This paper proposes a remote network emulation approach that utilizes a distributed server-based architecture in which local low-fidelity emulators provide real-time QoS predictions to distributed applications, coupled with a remote large scale high-fidelity simulator that continuously updates and calibrates the local low-fidelity emulators. Experimental results examining emulation results show that the remote network emulation system provides a promising approach to network emulation supporting accuracy and scale while meeting real-time constraints.