Introduction

This repository contains both the client and server side code for the ARPA-E-Sensor project HPDMobile, along with setup guides for configuring the Antsle 'antlets' (VM's). Each server is a raspberry pi with the following peripherals:

• Raspberry pi camera (photos)
• SPG30 Gas Sensor (TVOC and eCO2)
• DHT22 Sensor (Temp and RH)
• SPH0645LM4H Microphone (sound)
• VL53L1X Distance Sensor (meters)
• APDS-9301 Sensor (lux)

Glossary

1. antlet: For our understanding, this is essentially a virtual machine which is running on the Antsle. We have multiple antlets with different functionalities, used to isolate applications.
2. env_params: All data besides photos and sound (TVOC, eCO2, temp, RH, distance, light levels)
3. server = Raspberry pi: since each raspberry pi will be running a streaming video server (UV4L) and server to transfer the env_params, audio, and photos to the Antsle for central storage, it is referred to as a server throughout this project.
4. client: An antlet whose purpose is to connect to one of the servers, capture data points, and store locally on the antsle. There will be a client antlet for each server.

Server

Raspberry pi must be running Stretch OS. Both the camera and I2C must be enabled. The UV4L Streaming Server module must be running, follow steps here. Additionally, see the README in the server directory for more details. The servers will collect env_params at a specified read_interval (server.conf), timestamp the data, and store it in memory. They listen on a specific port listen_port, waiting for the client to request data. Upon request, data will be transferred over a TCP socket. If the client successfully receives and writes the data to the InfluxDB, the server will clear those readings from memory and begin collecting for the next interval.

Antsle

The Antsle will contain multiple antlets with different functions:

• 5-6 antlets will be used as the client antlets
• 1 antlet will be used as the USRP antlet

Influxdb Antlet

This antlet is running at 10.1.1.10 and is a debian - LXC. It contains the influx database. To install invluxdb on the antlet, copy the setup_scripts/antsle_influx_setup.sh into this antlet and run it at the command line. See the script for more details.

All env_params from the clients are being written to the hpd_mobile database running on this antlet. Although influxdb doesn't use tables (as in traditional relational databases), it uses what's called a 'measurement', which is essentially similar to a table. The 'measurement' used for this is also called 'env_params', which can be seen in the MyClient.influx_write method in client.py. To access the database:

1. ssh into this antlet
2. run $ influx at the CLI - now you will be at the influx CLI
3. run > use hpd_mobile
4. Now you can make similar queries as you would in a RDBMS. For instance > select * from env_params where server_id='S3'

Client Antlets

The client antlets will write env_params to the Influxdb Antlet. They will additionally write photos and audio data to a virtual drive mounted at /mnt/vdb/ (as specified by the client.conf img_audio_root parameter). 400 GB HDD has been allocated for the images and audio

Images

Images will be saved under /mnt/vdb/<server_id>/img/<date>/<%H:%M>. The server id will correspond to the id defined in the client.conf file. There should be 1,440 directories in the /date directory (one for every minute of the day). 60 photos should reside inside each of the /<%H:%M> directories, one for each second.

Audio

Audio will be saved under /mnt/vdb/<server_id>/audio/<date>/<%H:M>. The server id will correspond to the id defined in the client.conf file. There should be 3 wav files in each <%H:%M> directories, one for each minute.

USRP Antlet

Archive

# Test run
Before testing, you must:
- Have an influxdb running locally on port 8086 with an `hpd_mobile` table
- Check the client.conf file and change the `root` setting to a spot on your localhost to save images and audio.
    - Dev root: `/Users/corymosiman/hpd_mobile`
    - Prod root: `/mnt/vdb/`