Venkatesan N. Ekambaram

Quantified Traveler: Travel Feedback Meets the Cloud to Change Behavior

We describe the design and evaluation of a system named Quantified Traveler (QT). QT is a computational travel feedback system. Travel feedback is an established programmatic method whereby travelers record travel in diaries, and meet with a counselor who guides the user to alternate mode or trip decisions that are more sustainable or otherwise beneficial to society, while still meeting the subjects mobility needs. QT is a computation surrogate for the counselor. Since counselor costs can limit the size of travel feedback programs, a system such as QT at the low costs of cloud computing could dramatically increase scale, and thereby sustainable travel. QT uses an application (app) on the phone to collect travel data, a server in the cloud to process it into travel diaries, and then a personalized carbon, exercise, time, and cost footprint. The subject is able to see all of this information on the Web. We evaluate the system with 135 subjects to learn whether subjects will let us use their personal phones and data plans to build travel diaries, whether they actually use the website to look at their travel information, whether the design creates pro-environmental shifts in psychological variables measured by entry and exit surveys, and finally whether the revealed travel behavior records reduced driving. Before-and-after statistical analysis and the results from a structural equation model suggest that the results are a qualified success.

Circulant structures and graph signal processing

Linear shift-invariant processing of graph signals rests on circulant graphs and filters. The spatial features of circulant structures also permit shift-varying operations such as sampling. Their spectral features-as described by their Graph Fourier Transform profiles-enable novel multiscale signal processing systems and methods. To extend the reach of circulant structures, we present a method to decompose an arbitrary graph or filter into a combination of circulant structures. Our decomposition is analogous to resolving a linear time-varying system into a bank of linear time-invariant systems. As an application, we perform multiscale decomposition on temperature data spanning the continental United States.

Critically-sampled perfect-reconstruction spline-wavelet filterbanks for graph signals

Inspired by first-order spline wavelets in classical signal processing, we introduce two-channel (low-pass and high-pass), critically-sampled, perfect-reconstruction filterbanks for signals defined on circulant graphs, which accommodate linear shift-invariant filtering. We then generalize to filters that process signals defined on noncirculant graphs. We apply these filters, which can be tuned to approximate desired frequency responses, to signals defined on synthetic graphs and examine their performance.

Multiresolution graph signal processing via circulant structures

We use circulant structures to present a new framework for multiresolution analysis and processing of graph signals. Among the essential features of circulant graphs is that they accommodate fundamental signal processing operations, such as linear shift-invariant filtering, downsampling, upsampling, and reconstruction-features that offer substantial advantage. We design two-channel, critically-sampled, perfect-reconstruction, orthogonal lattice-filter structures to process signals defined on circulant graphs. To extend our reach to noncirculant graphs, we present a method to decompose a connected, undirected graph into a combination of circulant graphs. To evaluate our proposed framework, we offer examples of synthetic and real-world graph signal data and their multiscale decompositions.

Spline-Like Wavelet Filterbanks for Multiresolution Analysis of Graph-Structured Data

Multiresolution analysis is important for understanding graph signals, which represent graph-structured data. Wavelet filterbanks permit multiscale analysis and processing of graph signals-particularly, useful for harvesting large-scale data. Inspired by first-order spline wavelets in classical signal processing, we introduce two-channel (low-pass and high-pass) wavelet filterbanks for graph signals. This class of filterbanks boasts several useful properties, such as critical sampling, perfect reconstruction, and graph invariance. We consider an application in graph semi-supervised learning and propose a wavelet-regularized semi-supervised learning algorithm that is competitive for certain synthetic and real-world data.

Wavelet-regularized graph semi-supervised learning

Graph semi-supervised learning (GSSL) is a technique that uses a combination of labeled and unlabeled nodes on a graph to determine a classifier for new, incoming data. This problem can be analyzed through the lens of graph signal processing. In particular, the penalty functions used in the optimization formulation of standard GSSL algorithms can be interpreted as appropriately-defined filters in the Graph Fourier domain. We propose a wavelet-regularized semi-supervised learning algorithm using suitably-defined spline-like graph wavelets. These wavelets are critically-sampled, perfect-reconstruction basis representations, in contrast to much of the existing work proposing overcomplete representations. Critical sampling is essential for controlling the complexity in applications dealing with large scale datasets. We are also interested in understanding when wavelet-regularized approaches perform better than traditional Fourier-based regularizers. We compare the performance of our proposed spline-like, wavelet-regularized learning algorithm (as well as other existing graph wavelet designs) to some standard graph semi-supervised learning techniques on synthetic and real-world datasets.

Collaborative High-Accuracy Localization in Mobile Multipath Environments

We study the problem of high-accuracy localization of mobile nodes in a multipath-rich environment where submeter accuracy values are required. We employ a peer-to-peer framework where nodes can get pairwise multipath-degraded ranging estimates in local neighborhoods, with the multipath noise correlated across time. The challenge is to enable high-accuracy positioning under severe multipath conditions when the fraction of received signals corrupted by multiple paths is significant. Our contributions are twofold. We provide a practical distributed localization algorithm by invoking an analytical graphical model framework based on particle filtering, and we validate its potential for high-accuracy localization through simulations. In a practical dedicated short-range communication (DSRC) mobile simulation setup, we show that the algorithm can achieve errors of <; 1 m 90% of the time, even when the fraction of line-of-sight (LOS) signals is less than 35%. We also address design questions such as “how many anchors and what fraction of LOS measurements are needed to achieve a specified target accuracy?” by showing that the Cramer-Rao lower bound (CRLB) for localization can be expressed as a product of two factors: a scalar function that depends only on the parameters of the noise distribution and a matrix that depends only on the geometry of node locations and the underlying connectivity graph. A simplified expression is obtained that provides an insightful understanding of the bound and that helps deduce the scaling behavior of the estimation error as a function of the number of agents and anchors in the network.

Enhancing context awareness with activity recognition and radio fingerprinting

Within context-aware computing, there is a growing interest in linking localization technologies with activity recognition in a cooperative way. Existing research works in this field face two main difficulties: lack of accuracy in their solutions and/or sophisticated hardware requirements. To avoid these issues, we present a light-weight, low-cost and high-accuracy system for localization and activity recognition, based on a smart phone and a single off-the-shelf wireless accelerometer attached to the waist. We process the accelerometer signal with the wavelet transform to precisely recognize different activities and obtain the velocity of the gait in real time. Additionally, we leverage the capabilities of the smart phone to accurately estimate locations making use of a multimode approach for radio fingerprinting. Eventually, we combine location information with activity recognition, observing a 9% improvement in the accuracy with which some activities are recognized.

Non-line-of-sight localization using low-rank + sparse matrix decomposition

We consider the problem of estimating the locations of a set of points in a k-dimensional euclidean space given a subset of the pairwise distance measurements between the points. We focus on the case when some fraction of these measurements can be arbitrarily corrupted by large additive noise. This is motivated by applications like sensor networks, molecular conformation and manifold learning where the measurement process can induce large bias errors in some fraction of the distance measurements due to physical effects like multipath, spin-diffusion etc. Given the NP-completeness of the problem, we propose a convex relaxation that involves decomposing the partially observed matrix of distance measurements into low-rank and sparse components, wherein the low-rank component corresponds to the Euclidean Distance Matrix and the sparse component is a matrix of biases. Using recent results from the literature, we show that this convex relaxation yields the exact solution for the class of fixed radius random geometric graphs. We evaluate the performance of the algorithm on an experimental data set obtained from a network of 44 nodes in an indoor environment and show that the algorithm is robust to non-line-of-sight bias errors.

Linking Computer Vision with Off-the-Shelf Accelerometry through Kinetic Energy for Precise Localization

In this paper we propose the integration of computer vision with accelerometry in order to provide a precise localization solution. In terms of accelerometry, our approach makes use of a single off-the-shelf accelerometer on the waist to precisely obtain the velocity of the user. This allows us to calculate the kinetic energy of the person being tracked, and link the accelerometry data with the computer vision part of the system, where we employ segmentation of local regions of motion in the motion history image to estimate movement, and we leverage the number of pixels within the movement silhouettes as a metric accounting for the kinetic energy and the distance to the camera for the person being tracked. The fusion of the data from both technologies with a Kalman filter delivers an accuracy in the localization solution of up to 0.5 meters.