Energy is a resource bottleneck in wireless sensing networks (WSNs) relying on energy harvesting for their operations. This is pronounced in WSNs whose data is used for remote parameter estimation because only a subset of the measured information can be transmitted to the estimator. While much attention has been separately paid to communication schemes for energy-aware data transmission in WSNs under resource constraints and controlled parameter estimation, there has yet to emerge a censoring policy that minimizes the variance of a measured process’ estimated component parameters subject to realistic constraints imposed by the WSN. Consequently, this paper presents the derivation of an optimal event-based policy governing data collection and transmission that accounts for energy and data buffer sizes, stochastic models of harvested energy and event arrivals, value of information of measured data, and temporal death. The policy is optimal in the sense that it maximizes the information rate of transmitted data, thereby producing the best possible estimates of the process parameters using the modified maximum likelihood estimation given the system constraints. Experimental and simulation-based results reflect these objectives and illustrate that the framework is robust against significant uncertainty in the initial parameter estimates.

Panoptic segmentation is a complex full scene parsing task requiring simultaneous instance and semantic segmentation at high resolution. Current state-of-the-art approaches cannot run in real-time, and simplifying these architectures to improve efficiency severely degrades their accuracy. In this paper, we propose a new single-shot panoptic segmentation network that leverages dense detections and a global self-attention mechanism to operate in real-time with performance approaching the state of the art. We introduce a novel parameter-free mask construction method that substantially reduces computational complexity by efficiently reusing information from the object detection and semantic segmentation sub-tasks. The resulting network has a simple data flow that requires no feature map re-sampling, enabling significant hardware acceleration. Our experiments on the Cityscapes and COCO benchmarks show that our network works at 30 FPS on 1024x2048 resolution, trading a 3% relative performance degradation from the current state of the art for up to 440% faster inference.

Physical activities and social interactions are essential activities that ensure a healthy lifestyle. Public open spaces (POS), such as parks, plazas and greenways, are key environments that encourage those activities. To evaluate a POS, there is a need to study how humans use the facilities within it. However, traditional approaches to studying use of POS are manual and therefore time and labor intensive. They also may only provide qualitative insights. It is appealing to make use of surveillance cameras and to extract user-related information through computer vision. This paper proposes a proof-of-concept deep learning computer vision framework for measuring human activities quantitatively in POS and demonstrates a case study of the proposed framework using the Detroit Riverfront Conservancy (DRFC) surveillance camera network. A custom image dataset is presented to train the framework; the dataset includes 7826 fully annotated images collected from 18 cameras across the DRFC park space under various illumination conditions. Dataset analysis is also provided as well as a baseline model for one-step user localization and activity recognition. The mAP results are 77.5% for pedestrian detection and 81.6% for cyclist detection. Behavioral maps are autonomously generated by the framework to locate different POS users and the average error for behavioral localization is within 10 cm.

This paper proposes a three-tier algorithmic framework as the basis for the flexible design of data-driven structural health monitoring (SHM) systems. The three major functions of the SHM system, including data normalization, feature extraction, and hypothesis testing (HT), are mapped to the three layers of the framework. The first tier of the framework is devoted to data normalization. Machine learning (ML) methods are adopted to normalize available data sets by binning data sets to similar environmental and operational conditions (EOCs) of the system. Specifically, affinity propagation clustering is used to delineate data into groups of similar EOC. Once data are normalized by EOC, the second tier of the framework extracts features from the data to serve as condition parameters (CPs) for damage assessment. To ascertain the health state of the structure, the third tier of the framework is devoted to statistical analysis of the CP through HT. An intrinsic goal of the study is to explore the modularity of the three tier framework as a means of offering SHM system designers opportunity to explore and test different computational block sets at each layer to maximize the detection capability of the SHM system. Various realizations of the three-tier modular framework are presented and applied to acceleration and EOC data collected from an operational 3-kW wind turbine. In total, 354 data sets are collected from the turbine, including tower lateral accelerations in two orthogonal directions at six heights, wind speed and wind direction; 317 of the data sets correspond to the wind turbine in a healthy state and 37 with the wind turbine in a damage state. Using quantitative metrics derived from receiver operating characteristic (ROC) curves, the damage classification capabilities of the framework are validated and shown to accurately identify intentionally introduced damage in the turbine.

This special issue provides readers with a picture of the current state of the art in the structural health monitoring field while highlighting the new research avenues that are being aggressively explored.

This poster presents a distributed model-based fault detection algorithm which is based on local pair-wise verification. We first show that there exists a linear relationship between the outputs of any pair of sensors. Therefore, a network can be partitioned into sensor pairs, and the relationship between a pair of sensors can be modeled by a linear model. In addition to detecting general faults happened within a sensor pair, we developed an algorithm for identifying non-linearity type of fault without the use of reference sensors. Preliminary performance analysis shows that this scalable algorithm achieves high diagnosis accuracy. Communication power is also greatly reduced by the distributed nature of the algorithm.

This paper introduces a novel use of concepts from combinatorial group testing and Kalman filtering in detecting faulty sensors in a network when faults are relatively rare. By assigning sensors to specific groups and performing Kalman filter-based fault detection over these groups, we can obtain a small binary detection outcome, which can be decoded to reveal the fault state of all sensors in the network. Compared to existing methods, our algorithm achieves similar or better detection accuracy with fewer tests and thus lower computational complexity. We perform extensive numerical analysis using a set of real vibration data collected from the New Carquinez Bridge in California using an 18-sensor network mounted on the bridge.

In recent years, improved wireless technologies have enabled the low-cost deployment of large numbers of sensors for a wide range of monitoring applications. Because of the computational resources (processing capability, storage capacity, etc.) collocated with each sensor in a wireless network, it is often possible to perform advanced data analysis tasks autonomously and in-network, eliminating the need for the post-processing of sensor data. With new parallel algorithms being developed for in-network computation, it has become necessary to create a framework in which all of a wireless network's scarce resources (CPU time, wireless bandwidth, storage capacity, battery power, etc.) can be best utilized in the midst of competing computational requirements. In this study, a market-based method is developed to autonomously distribute these scarce network resources across various computational tasks with competing objectives and/or resource demands. This method is experimentally validated on a network of wireless sensing prototypes, where it is shown to be capable of Pareto-optimally allocating scarce network resources. Then, it is applied to the real-world problem of rupture detection in shipboard chilled water systems.

This poster presents a distributed reference-free fault detection algorithm which is based on local pair-wise verification. We show there exist a linear relationship between the output of any pair of sensors if the system excitations can be aggregated as a single system input. Using this relationship, faulty sensors suffering from sparse spike errors can be identified with high accuracy by our algorithm. An appealing feature of our method is that existence of reference sensors and knowledge of system input are not required. preliminary performance analysis shows that the algorithm is scalable, robust and able to detect most of the faults exist in the sensors. Communication power is also greatly reduced by the distributed nature of the algorithm.

The engineering community has recently begun to adopt wireless sensing technologies for use in many sensing applications. These low-cost sensors provide an optimal setting for dense sensing networks, and can make large amounts of sensor data available. Also, the computational power embedded within each sensing node allows a wireless network to interrogate data within the network and in real-time. In a monitoring situation, these capabilities can be leveraged to detect and locate changes in system properties that could be an early indication of malfunction within a complex physical system. In this paper, a distributed model updating technique is embedded within the computational core of a wireless sensing network. Using a novel distributed implementation of the simulated annealing method, an ad-hoc network of wireless sensing units can determine updated system properties by iteratively matching data derived from an analytical model of the system with collected sensor data. By comparing updated system properties with those obtained in a baseline state, a wireless sensing network can detect and locate changes in the system. Experimental verification of this technique is provided on a network of wireless sensor prototypes using simulated results from a Navy ship-board chilled water system monitored with a wireless sensor network.