Andrew Parker, PhD

Adaptive sampling algorithms, such as those used with mobile platforms to estimate 2D environmental phenomena, are traditionally point-based and designed to minimize the total number of point samples taken. However, in cases where the cost of sampling is negligible, such as with a photosynthetically active radiation (PAR) sensor, the energy cost is dominated by movement; in these cases, the goal should be to minimize the total length of travel. ... This dissertation presents an adaptive… Show more

Adaptive sampling algorithms, such as those used with mobile platforms to estimate 2D environmental phenomena, are traditionally point-based and designed to minimize the total number of point samples taken. However, in cases where the cost of sampling is negligible, such as with a photosynthetically active radiation (PAR) sensor, the energy cost is dominated by movement; in these cases, the goal should be to minimize the total length of travel. ... This dissertation presents an adaptive sampling algorithm that produces a field estimate whose mean squared error is less than that produced by a uniform raster scan of the same total travel length. The main problem is broken up into two related subproblems: first is how one generates field estimates conditioned on polygonal curve observations; second is how one selects a path observation given a field estimate. Two different Boolean models are studied: the Bombing model (one of the simplest Boolean random field models), and the polygonal random field (an extremely flexible example of a Boolean random field model). Path selection is restricted to polygonal curves (sequences of line segments). The approach for the two subproblems enjoy a similar formulation: the "solution" is a parameterized configuration (a Boolean field or a path); structured from an unknown number of components (i.e. is of variable dimension); a probability measure is defined over the space of valid configurations; reversible jump Markov chain Monte Carlo is used (employing Metropolis-Hastings-Green updates) to find a configuration that approximates either a mode or weighted average, which is the solution used for the path selection or field estimate problem respectively. The approach's performance is demonstrated against raster scan (for path selection) and bivariate linear interpolation (for the field estimate), and shown to be competitive in terms of the mean squared error. Show less

Imagers are an increasingly significant source of sensory observations about human activity and the urban environment. ImageScape is a software tool for processing, clustering, and browsing large sets of images. Implemented as a set of web services with an Adobe Flash-based user interface, it supports clustering by both image features and context tags, as well as re-tagging of images in the user interface. Though expected to be useful in many applications, ImageScape was designed as an analysis… Show more

Imagers are an increasingly significant source of sensory observations about human activity and the urban environment. ImageScape is a software tool for processing, clustering, and browsing large sets of images. Implemented as a set of web services with an Adobe Flash-based user interface, it supports clustering by both image features and context tags, as well as re-tagging of images in the user interface. Though expected to be useful in many applications, ImageScape was designed as an analysis component of DietSense, a software system under development at UCLA to support (1) the use of mobile devices for automatic multimedia documentation of dietary choices with just-in-time annotation, (2) efficient post facto review of captured media by participants and researchers, and (3) easy authoring and dissemination of the automatic data collection protocols. A pilot study, in which participants ran software that enabled their phones to autonomously capture images of their plates during mealtime, was conducted using an early prototype of the DietSense system, and the resulting image set used in the creation of ImageScape. ImageScape will support two kinds of users within the DietSense application: The participants in dietary studies will have the ability to easily audit their images, while the recipients of the images, health care professionals managing studies and performing analysis, will be able to rapidly browse and annotate large sets of images.
Show less

We envision the blossoming of sensing applications in an urban context, enabled by increasingly affordable and portable sensing hardware, and ubiquitous wireless access to communication infrastructure. In this paper, we describe the Partisans architecture, featuring infrastructure-supported selective data sharing and verification services. This effort represents an evolution of activity on embedded networked sensing from the scientific application space to applications in a space that raises… Show more

We envision the blossoming of sensing applications in an urban context, enabled by increasingly affordable and portable sensing hardware, and ubiquitous wireless access to communication infrastructure. In this paper, we describe the Partisans architecture, featuring infrastructure-supported selective data sharing and verification services. This effort represents an evolution of activity on embedded networked sensing from the scientific application space to applications in a space that raises novel issues in privacy, security, and interaction with the Internet. Show less

This paper introduces the concept of participatory sensing, which tasks everyday mobile devices, such as cellular phones, to form interactive, participatory sensor networks that enable public and
professional users to gather, analyze and share local knowledge. An initial architecture to enhance data credibility, quality, privacy and ‘shareability’ in such networks is described, as well as a
campaign application model that encompasses participation at personal, social and urban scales.… Show more

This paper introduces the concept of participatory sensing, which tasks everyday mobile devices, such as cellular phones, to form interactive, participatory sensor networks that enable public and
professional users to gather, analyze and share local knowledge. An initial architecture to enhance data credibility, quality, privacy and ‘shareability’ in such networks is described, as well as a
campaign application model that encompasses participation at personal, social and urban scales. Example applications are outlined in four areas: urban planning, public health, cultural identity and creative expression, and natural resource management. Show less

We describe Tinker, a high-level design tool that aids the exploration of the design space in sensor network applications. Tinker is targeted at applications that require real-time assignment of semantic meaning to data, rather than just data storage. Tinker lets users write simple programs, as if they were manipulating individual scalar values, and simulates those computations over continuous streams of sensor data. Tinker does not require (or allow) users to specify details such as routing… Show more

We describe Tinker, a high-level design tool that aids the exploration of the design space in sensor network applications. Tinker is targeted at applications that require real-time assignment of semantic meaning to data, rather than just data storage. Tinker lets users write simple programs, as if they were manipulating individual scalar values, and simulates those computations over continuous streams of sensor data. Tinker does not require (or allow) users to specify details such as routing algorithms or retransmission policies, freeing system designers to rapidly iterate among different broad designs before fleshing out details of the one that looks most promising. We demonstrate Tinker's use in the design and deployment of ElevatorNet, our distributed sensor application that retrofits buildings with per-floor displays of an elevator's position, determined using barometric altimetry. Show less