Experiment-Creation

Note

TODO: WORK IN PROGRESS

This Section describes how the user can utilize the shepherd-testbed. The parameters of all config-models are described in Shepherd-Core - Config-Models in detail. As most parameters have default-values the user only has to specify parameters that differ.

There are three options to describe an experiment:

YAML

datatype: Experiment
parameters:
  id: 4567
  name: meaningful Test-Name
  time_start: 2033-03-13 14:15:16
  target_configs:
  - target_IDs: [9, 10, 11]
    custom_IDs: [0, 1, 2]
    energy_env:
      name: SolarSunny
    virtual_source:
      name: diode+capacitor
    firmware1:
      name: nrf52_demo_rf
  - target_IDs:
      - 1
      - 2
      - 3
      - 4
    custom_IDs: [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17]
    energy_env:
      name: ThermoelectricWashingMachine
    virtual_source:
      name: BQ25570-Schmitt
      harvester:
        name: mppt_bq_thermoelectric
    firmware1:
      name: nrf52_demo_rf
    firmware2:
      name: msp430_deep_sleep

Source

Python

"""How to define an experiment:

- within python (shown in this example)
    - object-oriented data-models of
        - experiment
        - TargetConfig -> shared for group of targets
        - virtualSource -> defines energy environment and converters
    - sub-elements reusable
    - scriptable for range of experiments
    - check for plausibility right away
- as yaml (shown in experiment_from_yaml.yaml)
    - default file-format for storing meta-data (for shepherd)
    - minimal writing
    - easy to copy parts
    - submittable through web-interface

"""
from shepherd_core import TestbedClient
from shepherd_core.data_models.content import EnergyEnvironment
from shepherd_core.data_models.content import Firmware
from shepherd_core.data_models.content import VirtualHarvesterConfig
from shepherd_core.data_models.content import VirtualSourceConfig
from shepherd_core.data_models.experiment import Experiment
from shepherd_core.data_models.experiment import TargetConfig
from shepherd_core.data_models.task import TestbedTasks

# generate description for all parameters -> base for web-forms
Experiment.schema_to_file("experiment_schema.yaml")

# for online-queries the lib can be connected to the testbed-server
# NOTE: there are 3 states:
#    - unconnected -> demo-fixture is queried (locally)
#    - connected -> publicly available data is queried online
#    - logged in with token -> also private data is queried online
tb_client = TestbedClient()
do_connect = False

if do_connect:
    tb_client.connect()

# Defining an Experiment in Python
hrv = VirtualHarvesterConfig(name="mppt_bq_thermoelectric")

target_cfgs = [
    # first Instance similar to yaml-syntax
    TargetConfig(
        target_IDs=[9, 10, 11],
        custom_IDs=[0, 1, 2],
        energy_env={"name": "SolarSunny"},
        virtual_source={"name": "diode+capacitor"},
        firmware1={"name": "nrf52_demo_rf"},
    ),
    # second Instance fully object-oriented (recommended)
    TargetConfig(
        target_IDs=list(range(1, 5)),
        custom_IDs=list(range(7, 18)),  # note: longer list is OK
        energy_env=EnergyEnvironment(name="ThermoelectricWashingMachine"),
        virtual_source=VirtualSourceConfig(name="BQ25570-Schmitt", harvester=hrv),
        firmware1=Firmware(name="nrf52_demo_rf"),
        firmware2=Firmware(name="msp430_deep_sleep"),
    ),
]

xperi1 = Experiment(
    id="4567",
    name="meaningful Test-Name",
    time_start="2033-03-13 14:15:16",  # or: datetime.now() + timedelta(minutes=30)
    target_configs=target_cfgs,
)

# Safe, reload and compare content
xperi1.to_file("experiment_from_py.yaml", minimal=False)
xperi2 = Experiment.from_file("experiment_from_py.yaml")
print(f"xp1 hash: {xperi1.get_hash()}")
print(f"xp2 hash: {xperi2.get_hash()}")

# comparison to same config (in yaml) fails due to internal variables, BUT:
xperi3 = Experiment.from_file("experiment_from_yaml.yaml")
print(f"xp3 hash: {xperi3.get_hash()} (won't match)")

# Create a tasks-list for the testbed
tb_tasks2 = TestbedTasks.from_xp(xperi2)
tb_tasks2.to_file("experiment_tb_tasks.yaml")

# Comparison between task-Lists succeed (experiment-comparison failed)
tb_tasks3 = TestbedTasks.from_xp(xperi3)
print(f"tasks2 hash: {tb_tasks2.get_hash()}")
print(f"tasks3 hash: {tb_tasks3.get_hash()}")

Source