We were inspired by the Air Louisville project ([link] when we saw the impact it had on the patients symptoms, and were further intrigued when we saw the direct financial impact the program had on the cities healthcare infrastructure. We sought to expand upon their work with higher resolution data and a more effective data collection vector.

What it does

Our Dashboard Analytics are where the value happens for our users, think crime maps, but for air quality. By equipping our users with a neighborhood-resolution analysis of the air quality, we provide a financial incentive that empowers individuals, homeowners, and cities to become custodians of their environment. This dataset can also be provided to our partners via a REST API, which would allow us to provide more sophisticated services (Such as Machine Learning powered Predictive Models) built on the dataset to our future clients.

How we built it

This is all possible due to the Azure IoT Platform. The IoT Platform allows our MQTT Server based sensor network to bring in our data. Any MQTT Broker can do that though, the real value of Azure shines here with its seamless pipelines into Databases, Dashboard Analytics, and Machine Learning. This drastically improves our time to market in the short term, and provides benefits in the long term through the built-in dashboard analytics and ease of integration with other Azure Cloud services, for example the Azure Kubernetes Service we deployed our react app on!

Challenges we ran into

Our team did an excellent job adapting to the online format. We've all been working remotely for some time now, but there is still something difficult about establishing a connection with strangers through a video for the first time. Despite this, by the end of the hackathon we were enjoying the process together. The worst case scenario is that we'll all walk away with new friends, which is a win in my book. From a technology standpoint, we had difficulties implementing the Azure Cosmos pipeline as we had hoped, but fortunately were able to fall back on our familiarity with Microsoft SQL Server. We also spent the first ~2 weeks of the hackathon "sharpening our axe" by really focussing on our users and how we could have an impact on Microsoft's Sustainability goals. This leaft us with less time than we had hoped for the actual hacking, and we had to scale back the scope of our project by dropping the ML components of the project. In the future, we hope to implement TinyML on our sensors to improve signal processing, thereby improving the reliability of our low-cost hardware, as well as cloud based neural networks to predictively model air quality trends for cities and health systems.

Accomplishments that we're proud of

Implementing a system with complexity ranging from IoT Devices to Kubernetes Clusters with some Dashboard Analytics (and soon. Machine Learning) in between leaves one with quite a sense of accomplishment, even in a prototype scenario such as this. Our team did an excellent job supporting one another and taking advantage of our highly cross functional skill sets. We are confident that this concept is quite strong, and look forward to finding a city or other partner to explore wide-spread deployment of a sensor network of this nature. Our team found it particularly exciting to have an opportunity to not only potentially help Microsoft meet their sustainability goals (a worthy aim on its own), but to also improve human health and reduce healthcare costs at the same time. The prospect of a tertiary benefit in improving home values for homeowners who are responsible custodians of the air they live in is just the cherry on top.

What we learned

At the beginning of the hackathon all four of our team members had some level of experience working with Azure, however none of us had worked with the IoT Platform at all. This project gave us all familiarity with the advantages of Azure's IoT Platform over a simple MQTT Broker running locally, both in the short and long terms for future projects. It also has to be said that we found your Kubernetes getting started Documentation is much easier to follow than Google's. Oh the irony.

What's next for Air Quality Dashboard Analytics

We are currently working on establishing a partnership with the city of Boston to apply our sensor network to their well-maintained, and city-owned Bike network. We believe they are an ideal partner since they have a track record of caring about their ecological impact demonstrated by their million trees initiative, and they have an effective centralized infrastructure in their bike sharing program. These two characteristics make them poised to maximize their benefit. We've also discussed expanding on our device to use a blockchain like Ethereum. This would create an incentive structure where ordinary citizens may be interested in purchasing the device and carrying it with them to mine a cryptocurrency using the "proof-of-coverage" incentive structure explored by the helium network ([link]

Built With

Share this project: