Sensing Subsystem

Overview

Sensing Subsystem is a high level sensor framework inside the OS user space service layer. It is a framework focused on sensor fusion, client arbitration, sampling, timing, scheduling and sensor based power management.

Key concepts in Sensing Subsystem include physical sensor and virtual sensor objects, and a scheduling framework over sensor object relationships. Physical sensors do not depend on any other sensor objects for input, and will directly interact with existing zephyr sensor device drivers. Virtual sensors rely on other sensor objects (physical or virtual) as report inputs.

The sensing subsystem relies on Zephyr sensor device APIs (existing version or update in future) to leverage Zephyr’s large library of sensor device drivers (100+).

Use of the sensing subsystem is optional. Applications that only need to access simple sensors devices can use the Zephyr Sensors API directly.

Since the sensing subsystem is separated from device driver layer or kernel space and could support various customizations and sensor algorithms in user space with virtual sensor concepts. The existing sensor device driver can focus on low layer device side works, can keep simple as much as possible, just provide device HW abstraction and operations etc. This is very good for system stability.

The sensing subsystem is decoupled with any sensor expose/transfer protocols, the target is to support various up-layer frameworks and Applications with different sensor expose/transfer protocols, such as CHRE, HID sensors Applications, MQTT sensor Applications according different products requirements. Or even support multiple Applications with different up-layer sensor protocols at the same time with it’s multiple clients support design.

Sensing subsystem can help build a unified Zephyr sensing architecture for cross host OSes support and as well as IoT sensor solutions.

The diagram below illustrates how the Sensing Subsystem integrates with up-layer frameworks.

Configurability

• Reusable and configurable standalone subsystem.

• Based on Zephyr existing low-level Sensor API (reuse 100+ existing sensor device drivers)

• Provide Zephyr high-level Sensing Subsystem API for Applications.

• Separate option CHRE Sensor PAL Implementation module to support CHRE.

• Decoupled with any host link protocols, it’s Zephyr Application’s role to handle different protocols (MQTT, HID or Private, all configurable)

Main Features

• Scope

  • Focus on framework for sensor fusion, multiple clients, arbitration, data sampling, timing management and scheduling.

• Sensor Abstraction

  • Physical sensor: interacts with Zephyr sensor device drivers, focus on data collecting.

  • Virtual sensor: relies on other sensor(s), physical or virtual, focus on data fusion.

• Data Driven Model

  • Polling mode: periodical sampling rate

  • Interrupt mode: data ready, threshold interrupt etc.

• Scheduling

  • single thread main loop for all sensor objects sampling and process.

• Buffer Mode for Batching

• Configurable Via Device Tree

Below diagram shows the API position and scope:

Sensing Subsystem API is for Applications. Sensing Sensor API is for development sensors.

Major Flows

• Sensor Configuration Flow

• Sensor Data Flow

Sensor Types And Instance

The Sensing Subsystem supports multiple instances of the same sensor type, there’re two methods for Applications to identify and open an unique sensor instance:

• Enumerate all sensor instances

  sensing_get_sensors() returns all current board configuration supported sensor instances’ information in a sensing_sensor_info pointer array .

  Then Applications can use sensing_open_sensor() to open specific sensor instance for future accessing, configuration and receive sensor data etc.

  This method is suitable for supporting some up-layer frameworks like CHRE, HID which need to dynamically enumerate the underlying platform’s sensor instances.

• Open the sensor instance by devicetree node directly

  Applications can use sensing_open_sensor_by_dt() to open a sensor instance directly with sensor devicetree node identifier.

  For example:

sensing_open_sensor_by_dt(DEVICE_DT_GET(DT_NODELABEL(base_accel)), cb_list, handle);
sensing_open_sensor_by_dt(DEVICE_DT_GET(DT_CHOSEN(zephyr_sensing_base_accel)), cb_list, handle);

This method is useful and easy use for some simple Application which just want to access specific sensor(s).

Sensor type follows the HID standard sensor types definition.

See include/zephyr/sensing/sensing_sensor_types.h

Sensor Instance Handler

Clients using a sensing_sensor_handle_t type handler to handle a opened sensor instance, and all subsequent operations on this sensor instance need use this handler, such as set configurations, read sensor sample data, etc.

For a sensor instance, could have two kinds of clients: Application clients and Sensor clients.

Application clients can use sensing_open_sensor() to open a sensor instance and get it’s handler.

For Sensor clients, there is no open API for opening a reporter, because the client-report relationship is built at the sensor’s registration stage with devicetree.

The Sensing Subsystem will auto open and create handlers for client sensor to it’s reporter sensors. Sensor clients can get it’s reporters’ handlers via sensing_sensor_get_reporters().

Note

Sensors inside the Sensing Subsystem, the reporting relationship between them are all auto generated by Sensing Subsystem according devicetree definitions, handlers between client sensor and reporter sensors are auto created. Application(s) need to call sensing_open_sensor() to explicitly open the sensor instance.

Sensor Sample Value

• Data Structure

  Each sensor sample value defines as a common header + readings[] data structure, like sensing_sensor_value_3d_q31, sensing_sensor_value_q31, and sensing_sensor_value_uint32.

  The header definition sensing_sensor_value_header().

• Time Stamp

  Time stamp unit in sensing subsystem is micro seconds.

  The header defines a base_timestamp, and each element in the readings[] array defines timestamp_delta.

  The timestamp_delta is in relation to the previous readings (or the base_timestamp)

  For example:

  • timestamp of readings[0] is header.base_timestamp + readings[0].timestamp_delta.

  • timestamp of readings[1] is timestamp of readings[0] + readings[1].timestamp_delta.

  Since timestamp unit is micro seconds, the max timestamp_delta (uint32_t) is 4295 seconds.

  If a sensor has batched data where two consecutive readings differ by more than 4295 seconds, the sensing subsystem runtime will split them across multiple instances of the readings structure, and send multiple events.

  This concept is referred from CHRE Sensor API.

• Data Format

  Sensing Subsystem uses per sensor type defined data format structure, and support Q Format defined in include/zephyr/dsp/types.h for zdsp lib support.

  For example sensing_sensor_value_3d_q31 can be used by 3D IMU sensors like SENSING_SENSOR_TYPE_MOTION_ACCELEROMETER_3D, SENSING_SENSOR_TYPE_MOTION_UNCALIB_ACCELEROMETER_3D, and SENSING_SENSOR_TYPE_MOTION_GYROMETER_3D.

  sensing_sensor_value_uint32 can be used by SENSING_SENSOR_TYPE_LIGHT_AMBIENTLIGHT sensor,

  and sensing_sensor_value_q31 can be used by SENSING_SENSOR_TYPE_MOTION_HINGE_ANGLE sensor

  See include/zephyr/sensing/sensing_datatypes.h

Device Tree Configuration

Sensing subsystem using device tree to configuration all sensor instances and their properties, reporting relationships.

See the example samples/subsys/sensing/simple/boards/native_sim.overlay

API Reference

Sensor Types

Data Types

Sensing Subsystem API

Related code samples

• Sensing subsystemGet high-level sensor data in defined intervals.

Sensing Sensor API