Deokwoo Jung

Sensor node lifetime analysis: Models and tools

This article presents two lifetime models that describe two of the most common modes of operation of sensor nodes today, trigger-driven and duty-cycle driven. The models use a set of hardware parameters such as power consumption per task, state transition overheads, and communication cost to compute a nodes average lifetime for a given event arrival rate. Through comparison of the two models and a case study from a real camera sensor node design we show how the models can be applied to drive architectural decisions, compute energy budgets and duty-cycles, and to preform side-by-side comparison of different platforms. Based on our models we present a MATLAB Wireless Sensor Node Platform Lifetime Prediction and Simulation Package (MATSNL). This demonstrates the use of the models using sample applications drawn from existing sensor node measurements.

Estimating building consumption breakdowns using ON/OFF state sensing and incremental sub-meter deployment

This paper considers the problem of estimating the power breakdowns for the main appliances inside a building using a small number of power meters and the knowledge of the ON/OFF states of individual appliances. First we solve the breakdown estimation problem within a tree configuration using a single power meter and the knowledge of ON/OFF states and use the solution to derive an estimation quality metric. Using this metric, we then propose an algorithm for optimally placing additional power meters to increase the estimation certainty for individual appliances to the required level. The proposed solution is evaluated using real measurements, numerical simulations and by constructing a scaled down proof-of-concept prototype using binary sensors.

Model-based design exploration of wireless sensor node lifetimes

This paper presents two lifetime models that describe two of the most common modes of operation of sensor nodes today, trigger-driven and duty-cycle driven. The models use a set of hardware parameters such as power consumption per task, state transition overheads, and communication cost to compute a nodes average lifetime for a given event arrival rate. Through comparison of the two models and a case study from a real camera sensor node design we show how the models can be applied to drive architectural decisions, compute energy budgets and duty-cycles, and to preform side-by-side comparison of different platforms.

PEM-ID: Identifying people by gait-matching using cameras and wearable accelerometers

The ability to localize and identify multiple people is paramount to the inference of high-level activities for informed decision-making. In this paper, we describe the PEM-ID system, which uniquely identifies people tagged with accelerometer nodes in the video output of preinstalled infrastructure cameras. For this, we introduce a new distance measure between signals comprised of timestamps of gait landmarks, and utilize it to identify each tracked person from the video by pairing them with a wearable accelerometer node.

Identifying people in camera networks using wearable accelerometers

We propose a system to identify people in a sensor network. The system fuses motion information measured from wearable accelerometer nodes with motion traces of each person detected by a camera node. This allows people to be uniquely identified with the IDs the accelerometer-node that they wear, while their positions are measured using the cameras. The system can run in real time, with high precision and recall results. A prototype implementation using iMote2s with camera boards and wearable TI EZ430 nodes with accelerometer sensorboards is also described.

Towards Cooperative Localization of Wearable Sensors using Accelerometers and Cameras

This work describes a new approach for localizing people by cooperative sensor fusion of lightweight camera and wearable accelerometer measurements. We present the algorithm to identify people moving around as they are detected by cameras deployed in the infrastructure. The algorithm uses a correlation metric to develop an ID matching algorithm that can associate people in the scene to their global ID emitted from a wireless accelerometer sensor node worn on their belts. First we conduct a set of preliminary experiments to verify that the quantities of interest easily measurable by off-the-shelf components. We validate our metric and the performance of the proposed ID matching algorithm using simulations on testbed data that also includes a crowded scenario.

Recognizing activities from context and arm pose using finite state machines

We present an activity-recognition system for assisted living applications and smart homes. While existing systems tend to rely on expensive computation of comparatively largedimension data sets, ours leverages information from a small number of fundamentally different sensor measurements that provide context information pertaining the persons location, and action information by observing the motion of the body and arms. Camera nodes are placed on the ceiling to track people in the environment, and place them in the context of a building map where areas and objects of interest are premarked. Additionally, a single inertial sensor node is placed on the subjects arm to infer arm pose, heading and motion frequency using an accelerometer, gyroscope and magnetometer. These four measurements are parsed using a lightweight hierarchy of finite state machines, yielding recognition rates with high precision and recall values (0.92 and 0.93, respectively).

An Energy Efficiency Evaluation for Sensor Nodes with Multiple Processors, Radios and Sensors

This paper constructs a model for studying the energy efficiency of sensor node architectures featuring a pair of low-end, low-power processor and radio and a pair of high-end, energy efficient processor and radio. Such nodes can have a highly dynamic range of operation ranging from the collection of simple temperature measurements or motion detection all the way up to sophisticated signal processing of sensor data. Our model explores the energy efficiency tradeoffs related to the decision of which processor and which radio should be used for each task. For this we derive a general Semi-Markov Decision process model for maximizing the asymptotic lifetime of two alternate designs, one with dynamic and one with static interconnect. The resulting models are validated with simulation and are applied to the reported measurements from an existing platform with two radios and two processors. Our results show how to quantify the gains of such design with respect to the power consumption properties of each component. Furthermore, based on our power budget calculation, we conclude that the deisgn of reconfigurable interconnect between multiple processors and radios would result in efficiency gains despite the energy overhead such device may incur.

EnergyTrack: Sensor-Driven Energy Use Analysis System

Demand side management (DSM) has emerged as a promising way to balance the electrical grids demand and supply in an economical and environmentally friendly manner. For successful DSM, it is crucial to automate the analysis of building energy usage with respect to important factors that drive it, such as occupancy. In this paper, we present a sensor-driven energy use analysis system, EnergyTrack, that continuously analyzes, evaluates, and interprets building energy use in real-time. We develop an energy usage model in EnergyTrack that simultaneously considers device-specific energy consumption, occupancy changes, and occupant utility. We also design a low-cost occupancy estimation algorithm with a lightweight training requirement. The EnergyTrack testbed is implemented in a commercial building office space. Through this testbed, we demonstrate the performance of our occupancy estimation algorithm and the application of EnergyTrack in energy use analysis.

Tracking appliance usage information in residential settings using off-the-shelf low-frequency meters

Given the ongoing widespread deployment of low frequency electricity sub-metering devices at residential and commercial buildings, fine-grained usage information of end-loads can bring a new powerful sensing modality in Cyber-Physical Systems (CPS). Motivated by the opportunity, this paper describes an algorithm of estimating the ON/OFF sequences for typical household end-loads in close-to-real-time using an off-the-shelf power meter. Unlike previous algorithms that lacks in scalability to support diverse applications in CPS our algorithm is designed to provide control knobs to support various trade-offs between accuracy and computation load or delay to satisfy the different application requirements. We experimentally verify the proposed algorithm using a collection of home appliances. Our experiment result shows that our algorithm is able to detect ON/OFF sequences of 7 appliances nearly without error and 3 appliances with moderate error rate less than 6% among 12 typical household appliances.