Commissioning the SIMPaCT smart sensor network
Setting up a large smart sensor network is more complex and time-consuming than you might think. Smart City research lead at the UTS Institute for Sustainable Futures, Andrew Tovey, talks about the highs and lows of it.
Measuring the complex mosaic of microclimates and soil moisture zones within Bicentennial Park takes a lot of sensors. For SIMPaCT, that's 50 smart temperature and humidity sensing devices, 13 weather stations and over 200 smart soil moisture sensing devices. Getting these devices all active, deployed, sending reliable data and documented in detail is a mammoth task. It involves a great deal of careful planning, a whole lot of time installing, and a considerable amount of follow-up work to verify correct installation, take photographs, and record vital contextual information about each device deployment. Thrown in the mix are clerical errors and mismatch of records that inevitably arise between the six different organisations involved in the setup of the network. Then there are a host of technical challenges, with nearly 25% of initial device installations failing to send regular data updates. With only a handful of devices with physical faults (unavoidable with that many devices), the main issue here is radio connectivity.
A majority of devices in the SIMPaCT network send their data via a local area radio network and with so many hills, gullies, banks and dense tree cover across the park, many devices struggle to phone home. Sometimes these issues can be solved with some clever software tweaks, sent to devices remotely, that help to boost their signal and get them heard. Much of the time however, it means that the team is back out in the park, digging up devices or climbing ladders. So when I get tired of staring at spreadsheets containing literally thousands of manually entered metadata fields, I get to escape the office, grab my wellies and an umbrella, and head out to Bicentennial Park.
Is my sensor broken, or is it just deployed next to a leaking sprinkler head? Things you can only discover by putting on your wellies!
I'll take a minute to mention the rain - it's been one of the wettest years on record here in Sydney and the irony of developing a smart irrigation project designed to address drought and heat is not lost on me. I've been out in the park during some serious downpours, with the rain flying in sideways, lawns turned into small lakes, and the normal trickle of Boundary Creek turned into raging white water. Apart from making for rather bracing fieldwork, the poor weather has actually exacerbated our challenges with the network installation. Rain attenuates the radio signal that devices use to send their data updates, and wet tree canopies can significantly block it. Heavily saturated soil quiets the signal from soil moisture sensors buried just below the ground. On the plus side, we get to see the 'worst case scenario' for device deployments in more marginal locations. If a device still transmits effectively during all this wet weather then we can be confident that it will work reliably under any conditions. We can also identify devices that need moving more easily.
If all this sounds like a rather messy and imperfect technology solution, I should emphasise that yes, it is! We are working at the forefront of smart city technology applications, using low-cost devices to do something that they have never been used to do before. Before smart low-cost technology like this came along, the cost of installing 260 sensors transmitting live environmental data in a tight grid across a parkland like Bicentennial Park would have been utterly prohibitive, not to mention completely impractical. Previous technologies are huge in comparison, and require additional bulky support equipment for power and comms. For the first time, a large smart sensor network like the one we have built for SIMPaCT is affordable and practically achievable. That means that a project like SIMPaCT is now possible for the first time - we are literally building a world-first. For this reason, nothing is tried and tested. We work stuff out through trial and error and unfortunately deal with a lot of tedious trouble-shooting. The good news is that, step by step, I can see this thing coming together. And with each step closer to a working system, we learn more about how to do this thing, allowing us to share our learning with others. We're being pretty meticulous documenting everything and we're excited to see how this first trial at Sydney Olympic Park can become a foundation for future smart irrigation projects across the country.
Left: Overseeing contractors deploying a sensor in a tree on an island in the middle of the lake. Right: Overseeing the installation of soil moisture sensors.
For now though, it's back to the park. There's a soil moisture sensor deployed under a line of giant fig trees, but it’s not precisely where I thought it was. The trees probably interfered with the GPS coordinates that were recorded when it went it. If I look carefully, I should be able to find some disturbed soil and track it down...
I know that soil moisture sensor is around here somewhere... (spot the small plastic tube marker)