I’m working on constructing wireless sensor networks with the Forest Service and a handful of Research Universities across New England for use in forested ecosystems (soil, temp, humidity, irradiance, etc). Although I’m an ecosystem scientist and not an electrical engineer, there is a growing need to get portable low-cost sensors out for monitoring in the research community. Also, I grew up roughly 6 miles from Stroud in Unionville, PA! What a small world!
Great to hear about it – this is a board to bring together the ecosystem sciences and software/electronic to make it happen.
Be great to hear how you are going about defining and refining the sensor networks.
I’ve been designing remote instream water sensors. The challenges has been low cost environmental protection, solar powering, field access and types of wireless, sensors, programming and of course reliability.
The Mayfly and the courses that Stroud have been offering are some very detailed handson steps.
Some times with a budget commercial devices like Onset U30, RX3000… can work depending on sensors.
I’ve also been wondering how to fit something like the BeagleBoneBlack – https://sensorgnome.org/ possibly having the flexibility of python language https://github.com/stanleylio/fishie
powering becomes the real cost with these systems. So I’ve been working on reducing the power used so can reduce the total cost of powering.
Be interested to hear where you are heading 🙂
Hi there Neil,
Well right now we’re quite focused on a handful of important characteristics of forest ecosystems, predominantly air T, soil T, plant available moisture, and volumetric soil moisture. The general paradigm shift we see now is that single sensor nodes do not provide the spatial or temporal resolution for us to understand microclimate variability in response to change (particularly understory vs. open fields). However, these changes are critical for providing climate refugia for small organisms, for nutrient cycling, energy balance, and fuel loading when thinking about fires.
I’ve been working on design for a sensor network that mimics a few that have been put together before in other research forests. Generally this consists of interacting base/router/end nodes that communicate via 900Hz rf wherein each node collects the aforementioned variables. Right now we are testing the wide range of available sensors with respect to their accuracy – then we’ll need to balance the sheer number of sensors and possible communications protocols. Down the road there will be power management and routing things to think about, but we’re not quite there yet!
Exciting to see that there are others interested in these sensors and boards in the context of the environmental sciences out there!
For a leafy canopy 2.4Ghz gets absorbed by the water. 900Mhz gives a much range in a leafy canopy for lower power. The WiFi standard is evolving to include a “low data rate” in sub 1GHz – but no time lines for which silicon will have it first.
I’d be interested in what you find with specific sensors – it can be painful – however calibration and drift are real world issues.
I’ve posted on some sensors that I’ve used for accurate low water depth monitoring and acceptance tested across a diurnal range –
so good luck.
So don’t know if this is of interest, I’ve been brainstorming on problems of managing power and network configuration in the field. Assuming scale ability and ease of use is an objective – that is data needs to get back onto the internet – the IoT “internet of things” paradigm.
From what I’ve seen for the internet wireless network configuration needs to be plug’n’play – which is way difficult and doesn’t have a lot of commercial advantage to the various industries.
The plug’n’play is that its often difficult to figure out before hand exactly what will work for specific locations.
The current paradigm is its all planned out exactly and tested in the lab before getting to the field – but I would suggest for working “land managers” and biologists the need is for more flexibility for configuration in the field.
For field radios distance is the ability to gain height above ground. The loggers themselves though have to be accessible 2-5′ above ground. The radios and antennas have to be closely integrated – playing around with RF connectors in the field is a specific skill.
A way of gaining height and being plug’n’play for connecting back to the logger is USB – 15’/5m or RS232 -50’/15m. Possible parts for terrestrial radios
though any cable might need protection from gnawing critters depending on local conditions :).
USB for small size connectors and cause it includes power seems ideal – though limited to15′ .
For the logger USB host is critical – and while I’ve played with a number of configurations – I’m currently making the Beagle Bone Black/green/wireless (BBB) series work – which all have one USB Host A and micro OTG.
The Beagle Bone Green Wireless has 4 USB A and micro OTG. (may be a problem for really cold climates)
Other varieties of Beagle Bone Black are designed for industrial temperature range,
The issue with any Linux device is power.
Currently that powering seems to be costing ~$500 with a large 12V battery footprint and large 65W solar panel.
If the data collection paradigm can be event based (a rewrite of a lot sw) – possibly on a ‘event tick’ of 15minutes or greater, then there is the possibility of reducing the power consumed significantly.
So for the “15minute sampling paradigm” it seems to me the challenge is on how to reduce a solar powered configuration to 15W solar panel, LiIon Pack (6Ahr) with a target of 2weeks with no sunlight.
Forests with limited sunlight are special powering challenges. Possibly replaceable modular LiIon battery packs
For more distributed networks say using multiple Mayflys with 900Mz Xbee with an uplink through a central connector, the challenge is how to synchronize an event based network.
Well just some observations – if they are useful :). I’m working on a special power management/interfaces board to go with the BBB 🙂