Riverscout - A distributed envrionmental monitoring system

This README provides info on what Riverscout is, it’s architecture and how to set it up. For more info, refer to the PDF (Project-Report.pdf) which is an edited version of the final year report submitted along with the source code. The report goes into more about the 'how' and 'why' behind the project as well as an explanation of why this set of tools & technologies were used.

• Riverscout - A distributed envrionmental monitoring system
  • What Is Riverscout?
  • System Architecture
    • Gauge:
    • Backend Server:
  • Configuring the Riverscout Environment
  • Setting up & Connecting a Gauge
  • API Docs
    • Device Info
      • POST: /addOrUpdateDevice
      • DELETE: /deleteDeviceInfo
      • GET: /getInfoForDeviceID
    • Device Group
      • GET: /getAllDeviceGroups
      • GET: /findDeviceGroup
      • GET: /getDevicesInGroup
      • GET: /getDevicesInCountry
      • POST: /addOrUpdateDeviceGroup
      • DELETE: /deleteDeviceGroup
    • Device Types
      • GET: /getAllDeviceTypes
      • DELETE: /deleteDeviceType
      • POST: /addOrUpdateDeviceType
    • Sigfox Device Readings
      • POST: /addSigfoxDeviceReading
      • DELETE: /deleteOneSigfoxDeviceReading
      • DELETE: /deleteAllSigfoxDeviceReadings
      • GET: /getReadingsForDeviceID
    • Country Codes & Info
      • POST: /addOrUpdateCountry
      • GET: /getAllCountries
      • DELETE: /deleteCountry
  • Data Compression Scheme
    • Code - device side
    • Code - server side

What Is Riverscout?

Riverscout was my final year college project submitted as part of my degree. It aims to provide near real-time analytics on environmental phenomena such as water levels, water temperature and air quality to provide data for commercial and recreational users.

The idea was that low-power and cost effective 'gauges' could be deployed to an area to monitor certain environmental phenomena like air temperature, river levels and soil quality (to name a few).

The data is available through an API. Thespec is listed below. The API allows a user to query based on a country, a group within that country and finally a device within a selected group. The timeframe can be specified to only return the amount of data required.

Overall the project was a proof of concept and not launched in a commercial sense. This code + documentation is provided for reference only.

System Architecture

The project focused on two main areas, the device 'gauge' and a backend server platform for collecting and aggregating data. A front end application based on CoreUI Vue was planned but abandoned due to time constraints.

Gauge:

Riverscout uses a Pycom SiPy with a US-100 ultrasonic sensor and a DS18B20 temperature sensor. The SiPy is a internet enabled microcontroller that connects to the Sigfox network and can be programmed using a specialized variant of Python called MicroPython.

• For the device code, please visit the device repo:

Backend Server:

Riverscout is written in Javascript, uses oas-tools for middleware/API docs and stores it's data in MongoDB via Mongoose. The server is powered by NodeJS and is tailored towards a Linux/Unix like environment.

NOTE Riverscout was not tested in a Windows environment. While Node and MongoDB are OS independent, I cannot guarantee that the app will function correctly.

Configuring the Riverscout Environment

Required Software

The Riverscout environment was developed and tested on Ubuntu 18.04 LTS. It also needs the following installed to function:

• Node JS (select the .deb version of the latest LTS release). Setup instructions available here: https://github.com/nodesource/distributions#deb

• MongoDB (use version 4 or above for Ubuntu Bionic 18.04). Instructions available on MongoDB website https://docs.mongodb.com/manual/tutorial/install-mongodb-on-ubuntu/

Database configuration

By default MongoDB comes with access control disabled and outside traffic blocked. When we open it up to external traffic, access control must be enabled otherwise the database will be wide open for anyone to access! See https://www.theregister.co.uk/2017/01/09/mongodb/ for an idea.

The following instructions were based on Ian London’s article “How to connect to your remote MongoDB server”, https://ianlondon.github.io/blog/mongodb-auth/

To setup a new user and configure access control:

1. Start the mongo service if it is not already running ( sudo service mongod start)
2. Check the status of the mongod.service ( systemctl status mongod.service ) If the service is running, log into Mongo by typing mongo at the shell. This brings you into the MongoDB Shell.
3. Switch to the ‘riverscout’ database by typing use riverscout. If this does not exist, Mongo will create it for you.
4. Create a new user by typing:

 db.createUser({
    user: 'riverscout',
    pwd: 'riverscout',
    roles: [{ role: 'readWrite', db:'riverscout'}]
})

The above will create a user called ‘riverscout’ with the password ‘riverscout’ If you have specified a different username and/or password, please modify the mongoose.connect line in app.js to reflect these changes.

5. Now the new user has been added, type db.auth("riverscout","riverscout") This should return 1 which means your new user can successfully authenticate on the ‘riverscout’ db
6. Exit the Mongo shell by pressing Ctrl+D
7. Edit the file /etc/mongod.conf with your editor of choice

By default the server is only bound to localhost so to open it to the outside world,uncomment the net portion of the config file.

Bind the server to all interfaces:
# network interfaces
net:
  port: 27017 (default port)
  bindIp: 0.0.0.0  NOTE this needs to be explicitly declared as such in Mongo DB versions 4 + 

If the port is changed from the default, please modify the mongoose.connect line in app.js to include the custom port. See Mongoose API Docs on connection options

1. Enable access control by modifying the 'security' part as follows

security: (uncomment this)
  authorization: 'enabled' (add this)

Save this file and exit. Restart MongoDB by typing sudo service mongod restart Check the status. If MongoDB fails to start, it is probably an indentation issue with the config file. Refer to the log file and the output from systemctl.

Vagrant based development environment

Not necessary for cloud deployment or development. Use only if desired.

Riverscout can use Vagrant to provision and manage virtual machines for it's dev environment. To set up the development environment with Vagrant, perform the following steps

1. Install Vagrant (+ whatever dependencies it needs)
2. Install Virtualbox and its corresponding expansion pack VMWare has not been tested and would require some modification to the Vagrantfile
3. Clone the dev-environment repo and cd into it
4. If necessary, change the amount of RAM the VM receives by editing the vb.memory line in the Vagrantfile to a value that fits your system. Min recommended value is 1024MB
5. Start the Riverscout VM by typing vagrant up. This will download and set up the new VM. When this is done, login with vagrant ssh.

Setting up & Connecting a Gauge

To set up a Pycom SiPy to talk to Riverscout:

1. Follow the steps to set up the backend server on a cloud provider such as AWS or Google Cloud. Make sure to properly secure your instance with appropiate security groups!
2. Register your SiPy on the Sigfox Backend . Steps to do this can be found on Pycom's website
3. Once the device is registered, follow the guide on programming a gauge. This is necessary for programming the SiPy and connecting the sensors.
4. Go to the Sigfox Backend and select 'Device Type' in the top menu.
5. Click the link in the 'Name' column in the table of available devices. This will take you to a page showing device info.
6. Click the 'Callbacks' link in the left-hand sidebar.
7. To create a new callback, click the 'New' button (top right) and select 'Custom Callback'. Fill out the fields as follows:

• Type = DATA
• Direction = UPLINK
• Leave the 'Custom Payload Config' box blank.
• URL Pattern = http://<IPADDRESS>:8080/api/addSigfoxDeviceReading , where IPADDRESS is the IP address of your cloud instance.
• Use HTTP Method: POST
• Content type: application/json
• Body: {"rawHexString" : "{data}","sigfoxID" : "{device}","timestamp" : "{time}"}

Here is what the edit callback screen should look like.

Click OK to save the callback.

API Docs

The Riverscout API, as explained in the introduction, is structured as follows:

1. The user selects a country using a 'country code'. The list of codes can be found by calling the /getAllCountries route.
2. Once a country has been selected, you can query for a list of device 'groups' within that country by calling /getAllDeviceGroups with the countries' code.
3. To get data for one device, you need to then query for all the devices within the selected group with /getDevicesInGroup.
4. Using the 'deviceID' of the gauge in question, you can call /getReadingsForDeviceID which will return data for the device within the specified timeframe.

My reasoning for this was scale. For a small number of devices, it is okay to simply return a list of all device ID's in the system with every request. But when the number of devices increases, the amount of data being transferred from server to client increases. To minimise the amount of wasted data (and processing on the server end), the API is designed to allow granular selection of devices.

Also the grouping allows large dots to be shown on a country map when it is zoomed out. As the user zooms in, the groups closest to the centre of the map window disappear and are replaced with dots indicating the devices.

When a large area is selected, the user only sees the groups (black markers)

When the map is zoomed in on a specific area containing groups, the map now shows the devices in those groups (red markers)

Map image from OpenStreetMap

API Routes

The following is a brief description of each API route, broken into the various categories.

Device Info

These routes handle CRUD operations for a gauge/device. Device info is defined as it's install location or date, country code, hardware IDs whether it's active or not and so on. For getting readings from a device, use the /getReadingsForDeviceID route.

POST: /addOrUpdateDevice

Adds a new device (Sigfox/NBIoT) or if the device already exists, updates it's details

Value	Type	Description
displayName	string	Name to display in front end
gpsLong	number	GPS longitude of the gauge install location
gpsLat	number	GPS latitude of the gauge install location
countryCode	string	Uses the same codes as web domains. Eg: IE= Ireland
sigfoxID	string	Sigfox ID as generated by the Sigfox backend
installDate	string	Install date of device in UTC. Please use ISO 8601 format. Eg: 2018-09-09T14:00:00Z
replacementDate	string	UTC formatted date of the expected date the device is to be replaced. Use ISO 8601 format.
EOLDate	string	UTC format date of the time a device was actuallly replaced
reportingFreq	string	How often (in minutes) the device is expected to send up data
groupIDS	Array [string]	An array of the group IDs this device is in
activeStatus	boolean	Device active status
downlinkEnabled	boolean	Whether the device allows data to be sent to it or not.

DELETE: /deleteDeviceInfo

Deletes the selected device

Value	Type	Description
deviceID	String	Database ID of device to delete data for.

GET: /getInfoForDeviceID

Returns data for the specified device like it's location and group

Value	Type	Description
deviceID	String	Database ID of device to return data for.

Device Group

CRUD operations on device groups.

GET: /getAllDeviceGroups

Returns all the device groups for the specified country code

Value	Type	Description
countryCode	String	Return all groups matching this country.

GET: /findDeviceGroup

Returns a device group matching the specified name

Value	Type	Description
groupName	String	The name of the group to return

GET: /getDevicesInGroup

Returns the IDs of the devices in a group

Value	Type	Description
groupID	String	Return devices in a group with this ID.

GET: /getDevicesInCountry

Note: this route is not required and was added to facilitate testing. Use the above method when querying the API

Returns the IDs of all devices matching a specified country code

Value	Type	Description
countryCode	String	Return devices with this countryCode.

POST: /addOrUpdateDeviceGroup

Adds a group for devices

Value	Type	Description
groupLat	Number	GPS Latitude of the new group.
groupLong	Number	GPS Longitude of the new group
groupName	String	Name of new group. MUST be unique otherwise will trigger an update instead of an add.
countryCode	String	Two character code representing countries

DELETE: /deleteDeviceGroup

Deletes the selected device group

Value	Type	Description
deviceGroupID	String	Database ID of the group to remove

Device Types

CRUD operations on device 'types'. This was not implemented fully in the above code due to enum type issues with Mongoose. A device has a type which described it's primary role. A river level sensor would have type 'river' for example.

GET: /getAllDeviceTypes

Has no parameters, returns all the possible types of devices

DELETE: /deleteDeviceType

Deletes the selected device type

Value	Type	Description
deviceTypeID	String	Database ID of the type to remove

POST: /addOrUpdateDeviceType

Adds a new / updates an exising device type

Value	Type	Description
deviceTypeName	String. Min:1, Max: 30	The name to use for a new device type
deviceType	String (Default: river)	The type of this new device
deviceTypeDescription	String. Min:1, Max: 250	Describe the device characteristics here

Sigfox Device Readings

CRUD operations of sensor readings from Sigfox networked devices.

POST: /addSigfoxDeviceReading

Adds a reading for a Sigfox device. This is the route used by the Sigfox backend to add data.

Value	Type	Description
sigfoxID	String	The Sigfox ID for a device
rawHexString	String	The raw hex data coming from the Sigfox backend
timestamp	String	UNIX timestamp coming from the Sigfox backend

DELETE: /deleteOneSigfoxDeviceReading

Deletes a single device reading.

Value	Type	Description
readingID	String	Database ID of the reading to remove

DELETE: /deleteAllSigfoxDeviceReadings

Deletes all readings for a Sigfox device matching the specified ID

CAUTION This action is irreversable!

Value	Type	Description
deviceID	String	Database ID of the device to remove readings for

GET: /getReadingsForDeviceID

Returns readings for the specified device between the timestamps specified

Value	Type	Description
deviceID	String	The ID of the device to get data for
timestampGt	String	Fetch results greater than this timestamp. Timestamp is to be specified in ISO 8601 format in UTC timezone
timestampLt	String	Fetch results less than this timestamp. Timestamp is to be specified in ISO 8601 format in UTC timezone

Country Codes & Info

CRUD operations for country objects which contain a code and name.

POST: /addOrUpdateCountry

Adds a new country with a name and a country code. If the specified country exists, it will be updated instead.

Value	Type	Description
countryName	String	Country Name (MUST be unique or will trigger update)
countryCode	String	Code for the new country

GET: /getAllCountries

Returns all countries.

DELETE: /deleteCountry

Deletes the selected country

Value	Type	Description
countryID	String	Database ID of the country to remove

Data Compression Scheme

Due to the constrained nature of LPWAN technologies, there has to be a method of compressing data generated on the device before tramsmitting it to the backend. Riverscout uses a scheme derived from this example provided by Austin Spivey/Wia.io

Both parts are as follows:

Code - device side

The Micropython code for compressing the data is below:

struct.pack('f',float(waterTemp)) + bytes([waterLevel])

This converts the float value (with 2 decimal places) into a 32 bit IEEE 754 value (which is the binary representation). The float is now represented with 4 hexadecimal characters (8 bits) which is combined with the temperature value and sent to Sigfox.

Specifying ‘f’ as the first argument in struct.pack tells Python to store the value as a C float The values are stored as little endian (least significant bit first) as the ESP32 is a little-endian architecture (as reported by sys.byteorder)

Code - server side

The hex string is passed from Sigfox into the ‘parseSigfoxData’ decompression function. It takes the raw string, slices the first 8 bits into one variable called ‘waterTemp’ and slices the rest of the buffer into another variable called ‘waterLevel’ To parse the waterTemp variable, use the readFloatLE() NodeJS function specifying 0 as the offset. We need to specify little endian as the encoding because the values were originally encoded that way.

To parse the ‘waterLevel’ value, we can use the readUInt8() function.

The full code is available to view in controllers/sigfoxReadingController.js.