Skip to content

Trade off exploration

The trade-off exploration tool - as shown below - allows to automatically run simulations in a certain (user-defined) design space and retrieves the requested performance metrics from the module's logs.

Architecture

More specifically, the users specifies:

  • a base configuration (of each module)
  • an arbitrary number of design parameters (of any module) to explore1
  • an arbitrary number of performance metrics (of any module) to retrieve1

The trade-off exploration tool then

  • permutes the specified parameters
  • sets up the simulation core accordingly
  • runs all the simulations concurrently (i.e., using multiprocessing, to speed up the exploration process)
  • retrieves the requested performance metrics and provides them to the user

Example

In this example, the number of successful checkpoints is retrieved for different settings of the checkpointing voltage threshold and capacitances1.

base_config = {'harvester' : harvest_config,
                'load' : load_config,
                'capacitor' : capacitor_config,
                'converter' : converter_config,
                'sim_time' : 10}

params = {'load.v_checkpoint' : list(np.arange(3.3, vhigh, 0.01)),
          'cap.capacitance : [3300e-6, 5100e-6]'}

metrics = [{'module' : 'load', 'params' : ['num_CHECKPOINT_successful']}]

result = run_tradeoff_exploration(params, metrics, base_config)

  1. For more details, refer to the paper and the simulation in Simulations/get_gameboy_checkpoint_threshold.py. ↩↩↩