ucsdwcsng / loratestbed

A LoRa testbed to evaluate PHY and MAC

LoRa Testbed

An infrastructure-less, arduino-compatible LoRa testbed for PHY and MAC research. The user will simply provide a config file (that contains the experiment configuration details) to the testbed. The controller will communicate with the network to then return the results after the experiment has ended. See below for a high level overview.

Diagram 1: High level overview of the LoRa Testbed.

Run a simple experiment with an example configuration

poetry run python3 ./loratestbed/run_testbed.py -g /dev/ttyACM0 -c /dev/ttyACM1 --config ./configs/example.yaml

You will need to have device ID 33 and 26 active to run this successfully.

Configuration format

The configuration YAML file should necessarily have the following format/fields:

device_list: [33, 26] # List of device IDs
experiment_time_sec: 10 # Experiment time in seconds
offered_load_percent: 80 # Aggregate network offered load
packet_size_bytes: 16
mac_protocol: "csma" # MAC protocol used by the devices
packet_arrival_model: "poisson" # Packet generation model used by the devices.
# PHY layer parameters
transmit_SF: "SF8" # Spreading factor
receive_SF: "SF8"
transmit_BW: "BW125" # Bandwidth: "BW125", "BW250", "BW500
receive_BW: "BW125"
transmit_CR: "CR_4_8" # Code rate
receive_CR: "CR_4_8"

Setup and installation

Setting up the testbed

Click here to go to the documentation on flashing various components in the testbed and bringing them up.

Host software

Install poetry

pip install -U poetry

Navigate to the root directory of this git repository. Run tests:

poetry run pytest

Install:

poetry install

loratestbed's Issues

Testbed does not handle cases where some devices consistently fail to send packets

There could be situations (where offered load is too high, or when experiment time is too small), when packets from one device are consistently masked by other packets from other devices (capture effect).

Add timestamps to the Gateway packet trace

README and Documentation

A lot of work needs to be put into readme and documentation

• Link/include video tutorial
• Describe result dataframe and how to use it to understand/plot metrics

CRC Bit might not be correctly accessed

LMIC.sysname_crc_err might not be returning the true value of the CRC bit. It seems to declare no error for packets that are erroneous.

193100000000000000000000F06D0619, 1, 0, 0 # this
1A2F0000000000000000000000000000, -3, 25, 0
1A300000000000000000000000000000, 0, 24, 0
1D320000000000000000000000000000, -8, 27, 0
1A310000000000000000000000000000, -3, 28, 0
1A320000000000000000000000000000, 1, 24, 0
1C320000000000000000000000000000, 0, 3, 0
1A330000000000000000000000000000, -2, 14, 0
1C340000000000000000000000000000, -8, 28, 0
1A340000000000000000000000000000, -2, 25, 0
19340000000000000000000000000000, -17, 28, 0
1C350000000000000000000000000000, -8, 26, 0
1C360000000000000000000000000000, -8, 23, 0
1A350000000000000000000000000000, 3, 10, 0
1C370000000000000000000000000000, -8, 23, 0
22310000000000000000000000000000, -3, 18, 0
1C380000000000000000000000000000, -8, 25, 0
1C390000000000000000000000000000, -9, 24, 0
1936000000000000008000007061E29E, -2, 3, 0 # This

Clean up top-level configuration parameters

Top level configuration should be able to:

• Change CSMA parameters
• Set offered load instead of mean inter-packet arrival time
• Tx and Rx PHY parameters should be set from the same place

Metrics computation uses hard-coded parameters

packet_duration = 0.126  # 126 ms
    packet_bytes = 16  # 16 bytes
    no_bits_in_byte = 8
    logger.warning(
        f"Using hard-coded values for packet duration: {packet_duration}, packet bytes {packet_bytes}"
    )