Commercial Internet of Things (IoT) sensors enable continuous data collection that benefits exposomic studies. The Exposure Health Informatics Ecosystem (EHIE) is one such sensor-based informatics platform for performing multiple simultaneous exposomic studies. It captures data from networks of sensors designed to record air quality in homes of the study’s participants and neighboring areas. In such cases where sensors are continually streaming data, it is crucial to monitor, in real time, the operational status of the network and record possible anomalies. Data collected by these sensors is only useful if it is free of errors. Therefore, maintaining the proper integrity of devices requires the capture of all deployment events that can cause anomalies. Tracking faults by recording system metadata is a difficult task, and we need a mechanism to capture the trajectories of devices within and across studies, systematically capture metadata of deployed version, and assign appropriate provenance to data recorded from each sensor. In this paper, we propose the use of a permissioned blockchain to manage the metadata and connect seemingly unrelated changes to create a trajectory of events that could result in the errors we observe. We implement a preliminary version of our blockchain solution in Hyperledger Fabric to help track errors in such a volatile setup. We also highlight how the properties of blockchain fulfill the essential needs for a metadata management solution needed in our case study.

The health hazard of air pollution in developing countries poses a significant threat of cardiovascular, respiratory, and other diseases. Ulaanbaatar, Mongolia is among cities with the worst polluted air in the world due to the use of coal as the primary heating source in the traditional Mongolian gers where most of the local population resides. Humanitarian groups are looking for ways to improve air quality, but are unable to measure the effects of their solutions. We build a low-cost air quality sensor that can upload data in real-time in remote locations. This newly developed sensor allows for _real-time_ air quality monitoring and tracking that was not possible before in such locations. We present the implementation and deployment of this system and share experiences and lessons learned from deploying the sensors in such a unique location.

We design and build a protocol called on-off noise power communication (ONPC), which modifies the software in commodity packet radios to allow communication, independent of their standard protocol, at a very slow rate at long range. To achieve this long range, we use the transmitter as an RF power source that can be on or off if it does or does not send a packet, respectively, and a receiver that repeatedly measures the noise and interference power level. We use spread spectrum techniques on top of the basic on/off mechanism to overcome the interference caused by other devices' channel access to provide long ranges at a much lower data rate. We implement the protocol on top of commodity WiFi hardware. We discuss our design and how we overcome key challenges such as non-stationary interference, carrier sensing and hardware timing delays. We test ONPC in several situations to show that it achieves significantly longer range than standard WiFi.

We design and build EpiFi, a novel architecture for in-home sensor networks which allows epidemiologists to easily design and deploy exposure sensing systems in homes. We work collaboratively with pediatric asthma researchers to design multiple studies and deploy EpiFi in homes. Here, we report on experiences from two years of deployments in 15 homes, of two different types of studies, including many deployments continuously monitored over the past 11 months. Based on lessons learned from these deployments and researchers, we develop a new mechanism for sensors to bootstrap their connectivity to a subject's home WiFi router and implement data reliability mechanisms to minimize loss in the network through a long-term deployment.

Air quality is important, varies across time and space, and is largely invisible. Pioneering past work deploying air quality monitors in residential environments found that study participants improved their awareness of and engagement with air quality. However, these systems fielded a single monitor and did not support user-specified annotations, inhibiting their utility. We developed MAAV– a system to Measure Air quality, Annotate data streams, and Visualize real-time PM2.5 levels – to explore how participants engage with an air quality system addressing these challenges. MAAV supports collecting data from multiple air quality monitors, annotating that data through multiple modalities, and sending text message prompts when it detects a PM2.5 spike. MAAV also features an interactive tablet interface for displaying measurement data and annotations. Through six long-term field deployments (20-47 weeks, mean 37.7 weeks), participants found these system features important for understanding the air quality in and around their homes. Participants gained new insights from between-monitor comparisons, reflected on past PM2.5 spikes with the help of their annotations, and adapted their system usage as they familiarized themselves with their air quality data and MAAV. These results yield important insights for designing residential sensing systems that integrate into users’ everyday lives.

When several internet-of-things devices are required to be installed in a smart home, significant effort is required to provide each device with the association information for the home's wireless router. We design and build a novel protocol called Secure Transfer of Association Protocol (STRAP), which securely bootstraps connectivity between a set of deployed WiFi devices and a home's wireless router. We show that STRAP works in a variety of environments and is faster than conventional methods for connecting WiFi devices to home wireless routers.