AWR1443BOOST: Phase stability across multiple chirps and frames

Phase shift will make it difficult to measure micro-motion. Therefore, I wonder how to improve the phase stability across multiple chirps (i.e., keep strictly coherent)? What is the reason for this shift or unstability? In additon, the phase shift may occur with different frames at the same time.

I use AWR1443Boost +DCA1000EVM caputer the raw radar data with mmWave Studio, and the monitoring object is static. From the studio software, we can see the phase stability performance for multiple chirps, which show obvious phase shift. Please see the attchment (No of chirp loops:128, chirp duration time :80 usphase stability across multiple chirps.pdf).