sensor_smoother.c

A small collection of sensor smoothing functions designed for embedded devices and other resource-constrained environments.

Pull requests and contributions are welcome!

Design Considerations

• Minimal runtime checks are enabled by default to catch common misconfigurations (e.g., null buffers or invalid alpha).
• Runtime checks can be disabled via compile-time flags to eliminate overhead when inputs are guaranteed valid.
• The simple moving average requires the user to supply and manage the buffer, giving flexibility in buffer sizing and memory management.
• The exponential moving average maintains internal state and requires setting a smoothing factor alpha between 0 and 1 (exclusive).
• No dynamic memory allocation used.
• The library uses float (single precision) as it is the most common size for sensor data in embedded platforms.
  • Users needing double precision can modify the source code accordingly, and contributions to support this are welcome.

Usage

Simple Moving Average

The simple moving average function maintains a circular buffer of recent input values and returns their average. The user must allocate and manage this buffer.

sensor_smoother_simple_moving_average_t sma_state = {0};
sma_state.buffer = buffer; // pre-allocated float array
sma_state.buffer_size = sizeof(buffer) / sizeof(buffer[0]);

float smoothed_value = sensor_smoother_simple_moving_average(&sma_state, input);

The function performs runtime validation by default:

• Returns the input value unchanged if the buffer is null or buffer size is zero.
  • Use this flag DISABLE_SENSOR_SMOOTHER_SIMPLE_MOVING_AVERAGE_CHECKS to disable null buffer and zero-size checks in the simple moving average function for performance reasons.

Exponential Moving Average

The exponential moving average uses a smoothing factor alpha between 0 (exclusive) and 1 (exclusive) to weight new input values. Smaller alpha values smooth more aggressively.

sensor_smoother_exponential_moving_average_t ema_state = {0};
ema_state.alpha = 0.1f;  // Smoothing factor: 0 < alpha < 1

float smoothed_value = sensor_smoother_exponential_moving_average(&ema_state, input);

The function performs runtime validation by default:

• Returns the input value unchanged if alpha is not strictly between 0 and 1.
  • Use this flag DISABLE_SENSOR_SMOOTHER_EXPONENTIAL_MOVING_AVERAGE_CHECKS to disable alpha range checks in the exponential moving average function for performance reasons.

Installation Via Clib

You can install this library using clib:

clib install mofosyne/sensor_smoother.c

This will download sensor_smoother.c and sensor_smoother.h into your project’s deps/ directory that you can then include into your codebase.