Concept for Python API

• lean on interface of established entry-level projects (i.e. wiring, arduino)

• submodules as classes ⇾ basic configuration at instantiation

• lib could be compiled for speed

• do we need exact timestamps for start and stop or is a global trigger enough

• hdf5 for fast storage, data can be segmented into several files, and written incrementally

  • later converters for csv, saleae logic-analyzer-format would be nice

• try to separate between scenario and measurement (reuse-ability)

• individualize target firmware by manipulating ELF-file: tx power and ID/softMAC as variable,

• modules

  • harvesting-emulator: energy-trace(s), sampling-rate, playback (mirrored, loop)

    • or skip emulator by using constant voltage module

    • traces are file based, there will be a default-folder with pre-selected ones (already on beagle)

  • harvesting-recorder: save-path, capture-duration, start-timestamp, sampling-rate

    • includes current-recorder and switch for dummy-load

  • dc-emulation-modul: converter model parameter, capacitor parameters, pre-charge …

    • low-level model in PRU

  • target-module: path for trace-dumps,

    • submodules: firmware (+addressing by firmware-manipulation), uart, gpios

    • allow to disconnect all iO (document in logs with timestamp + Event)

• bidirectional GPIO usage

  • allow to pass thread-function to nodes that handle reactions

• start of measurement should be triggered by absolute timestamp and marks T=0

  • what if the absolute time was missed? ⇾ synced start whenever start-signal comes, should be enough

  • after that it seems easier to use relative time increments for controlling submodules

  • interact with cron-jobs or other linux-scheduler

  • currently implemented: absolute time start with scheduler, node needs no control after config