Encrypted cloud-based MPC for linear systems with input constraints

Abstract We present a novel cloud-based MPC scheme that is based on a recently proposed real-time proximal gradient method. The cloud-based implementation requires sensitive data (e.g., system states) to be transmitted via public networks and to be processed in the cloud. We guarantee privacy of the data throughout the control-loop by encrypting the control scheme using (partial) homomorphic encryption. The resulting encrypted MPC computes encrypted predictive control actions based on encrypted system states (without intermediate decryptions).