Notes Phase 1 Integration Issues

Remote Sensor

• Remove the time based sync strategy between the Remote Sensor and Data Aggregator.
  • This was a case of pre-mature optimization
  • Replace with a slow (approx 1 Hz) advertising and measure the power consumption.
  • This would significantly simplify the upload processing and give us additional flexiblity in sampling rate (i.e. we could upload more frequently).
• Solve the time drift problem.
  • Try to configure the sleep driver to use the PRORTC peripheral. This is supposed to work on the lastest 2.xx SDK. This would allow us to return to using the RTC and we have the driver for it already. This has the advantage that the RTC can survive reset with the right control bit in the reset controller.
  • If the first try doesn't work, we could use the LETIMER peripheral for micca and leave the sleep driver to the BLE stack.
• Synchronize sampling to known time values, i.e. rather than starting sampling at an arbitrary time, align the start with the nearest hour. This is much like what is currently done with the upload times.
• Modify the BLE service configuration to better support transferring multiple Sensor Groups.
• Incorporate a sensor type attribute in the Environmental Sensor class that can be used to map the sensor to its properties and can be sent with the sample data information to define the sensor type which took the measurement.

Data Aggregator

Remote Sensor Upload Domain

• During service discovery, obtain the values of all the read only characteristics and cache them. This would simplify all the processing associated with getting those values when needed.
• Many of the values that need to be uploaded fall into this category. Could also remove the Profile Hash Read operation.

Aggregation Management Domain

• Add the concept that there are multiple data sets associated with a Reporting Sensor.
• The Reporting Spec class is problematic but we still need to be able to control the sampling interval for each sample set independently.
• Most of the Reporting Spec attributes go away with the removal of the sync scheme.

Data Accessor Offload Domain

• Communication protocol must be changed to handle multiple sample sets within a sensor.
  • Add additional message types to:
    • Meta-data about the sample set:
      • sensor serial number
      • sample interval
      • set of sensor type codes
    • Sample data values:
      • timestamp
      • set of sensor values
  • The above being needed for each sample set within a Reporting Sensor.
  • This will cause some interfaces changes for how a Service Session operates.

Data Accessor

• Need addition model to demonstrate how Time Series data is retrieved and mapped to the sensors which provided the measurements.
• Started working on a "one big domain" model.
• Additional software architecture work is needed. The current experiments "inverts" the usual architectural mapping. I'm looking at what type of mapping this implies for the domain model to translation.