Motion sensing with for examples MEMS technology is the most effective way of doing this. With multiple axis of motion data and from this you can derive the impact on the optical signal and compensate for this to reject the artifacts. There are people claiming that you can detect also motion with different wavelengths of light and use the differential signal for Heart rate detection and the common optical signal to reject motion. Personally I am not in favorite of this, as you don’t know if the wavelength used for rejection, is not somewhat influenced by the heart rate, and also two LEDs in your system, take 2x the power. This is much more compared to the ADxL362 which consumes just a few uA.
While designing a wearable, make sure the device is light. The more mass you have, the bigger the motion impact will be to your actual signal. Also design the mechanics in such a way, that the device doesn’t move very easily.
