Research

You've now heard about the semester project in Lecture 2: Sentimet. While we form teams, is time to do some research in preparation for your first team meeting in Lab02.

Requirements, specifications, and systems oh my

We have several interconnected aspects of design that we are trying to untangle:

We are trying to develop specifications for our system, which will enable us to engage in detailed engineering design. However, there is not a single-valued mapping from the requirements to their specifications. Further, it's possible that some requirements are not feasible. Finally, while the specifications will influence the system archictecture, the architecture can also influence the specifications. We have a bit of a head start this year, because we can build on what teams developed last year. Working mostly independently, they settled on similar system architectures:

Some key features:

• Weather is measured using a RH/T sensor.
• Occupancy is estimated by sniffing WiFi MAC addresses (one team also explored Bluetooth sniffing).
• Use of MAC addresses can be privacy-preserving as long as those addresses are not stored long-term.
• The system is controlled by an ESP32C3 module, which includes an FCC-certified WiFi radio.
• Data is sent back to the server over a FCC-certified cellular link, because of concern that utilizing the bus WiFi would not be reliable.
• Energy is managed using a power management IC (the MCP73871) that was connected to a battery, and recharged using solar energy (collected from a solar panel).
• The industrial design is comprised of a weatherproof case (purchased or 3D Printed) that can be mounted on a standard bus stop pole, and oriented to face south. A solar panel was mounted atop the panel.
• Teams used their own Raspberry Pi servers, storing data in databases and presenting via simple plots, some of which integrated with data from Swift.ly.

This was great progress, but not surprisingly the teams did not achieve everything they wanted. first pass. Here are some of the outstanding issues:

• UV-resistant polymers are better suited for objects that are going to be exposed to sunlight.
• Weather resistance of the units was not fully tested.
• Estimating occupancy via MAC sniffing showed promise but was not completely debugged.
• Teams did not get to run their systems long term (days to weeks) to tease out rare bugs.
• Teams did not fully implement energy management and demonstrate long-term battery-powered operation including multiple charge/discharge cycles.
• Electronics BOM (Bill of Materials) for these systems was somewhat high (~$100+), suggesting an overall COGS in the $200-300+ range. Lower COGS (Cost of Goods Sold) would allow these systems to be more ubiqitiuously deployed.
• This year's project also should measure sun exposure and surface temperature.
• Teams did not ensure that their systems could withstand shipping, and could be installed and set up by a "regular person".
• We need a variety of mounting options this year. Not just the standard bus stop pole, but also a tripod, and trying to be as "universal" as possible in mounting will be helpful.

We encourage you to start from these system diagrams, but it's worth researching some of the outstanding issues and how they might impact your system diagram.

For this exercise, choose one of the issues from the list below to research. You'll upload your findings to your team Google Drive in Lab02, and discuss with your team at the first team meeting.

Occupancy sensing

As noted above, teams ended up using MAC address sniffing. This is a well-established way to count the number of devices in range of your WiFi or Bluetooth radio, and from that estimate number of people and how long they dwell within range.

However, MAC address sniffing has some subleties. A few to consider:

• Should you sense WiFi, BLE, or both?
• How do you go from number of devices to number of people?
• MAC address beacons come with an estimate of signal strength. How do you use that to make sure you are only measuring devices that are "local" and not far away?

Luckily, MAC address sniffing is pretty common. Here are some resources:

• https://matthew.science/posts/occupancy/
• https://github.com/cyberman54/ESP32-Paxcounter/tree/master
• https://www.bluefox.io/mission
• https://github.com/gyf304/occupancy-sniffer

There are many others online.

Research the answers to some of these questions. Write them up in a google doc or other shareable document.

Sensing sun and surface

The MIT Office of Sustainability is interested not only in the air temperature, but the temperature of surfaces around (like the ground), as well as sun exposure (is the area shaded, partially sunny, and so on). This will require adding some sensors to the system.

Sun This is a bit subtler than it sounds. An object in a large open field would get "full sun", but the duration will depend on the time of year (unless we're at the equator...). In addition, if the day is cloudy, does that mean there is no sun exposure? Probably not.

So really what we want is a measure of the intensity of sunlight at the location of the sensor, because that sunlight will in turn influence the local air and surface temperatures.

Luckily, there are lots of sensors that can do this. Your phone, for instance, adjusts the screen brightness for the ambient light (brighter in sun, dimmer indoors). Look into ambient light sensors. How do they work? How are they connected to the rest of the system (UART, I2C, etc.)? What voltage and current do they need? How much do they cost?

Surface temperature The surfaces in a space can be hotter (or cooler) than the air. For example, you've surely had the experience of stepping barefoot on concrete in the sun and burning your feet. It gets hot! Or cooking eggs on the hood of a car.

But we will be mounting Sentimet on a pole, or on a tripod. So how do we measure the temperature of the ground that's at a distance? With a non-contact temperature sensor. Again, this is an entire class of sensors, and the same questions arise as to the ambient light sensor.

Industrial design

Teams used either 3D-printed or purchased enclosures. Purchased enclosures have the benefit that they can have built-in UV resistance, and appropriate ingress protection, at least until you drill holes in them! However, they come in limited shapes and sizes, so mounting them can be more challenging.

3D-printed parts have the benefit of being much more customizable, giving a more professional look. However, UV resistance is an issue, and the teams need to design the enclosure to make sure it can withstand the environment.

As you consider an enclosure for this year, some questions to look into:

• Are there 3D printers on campus or available as commercial services that will print plastics appropriate for outdoor use?
• If making your own enclosure, what are strategies for making it weathertight, and how do you test that?
• What are some options for purchasing enclosures? Some starting points are here, here, here, here, and of course Amazon, Aliexpress. Of note, bringing wires/cables into/out-of the enclosure in a weathertight way is possible using "cable glands".

What's next?

We'll be going over your findings as a team in Lab02.