In this chapter, we focus on encrypted model predictive control (MPC) implemented in a single cloud. In general, encrypted control enables confidential controller evaluations in networked control systems. Technically, an encrypted controller is amodified control algorithm that is capable of computing encrypted control actions based on encrypted system states without intermediate decryptions. Encrypted control can, for example, be realized using homomorphic encryption that allows simple mathematical operations to be carried out on encrypted data. However, encrypting optimization-based control schemes such as MPC is non-trivial. Against this background, the contribution of the chapter is twofold. First, we summarize and unify two existing encrypted MPCs using the additively homomorphic Paillier cryptosystem. Second, we present a novel encrypted MPC based on real-time iterations of the alternating direction method of multipliers (ADMM). We theoretically and experimentally compare the three approaches and highlight unique features of the new scheme.
CITATION STYLE
Darup, M. S. (2019). Encrypted model predictive control in the cloud. In Privacy in Dynamical Systems (pp. 231–265). Springer Singapore. https://doi.org/10.1007/978-981-15-0493-8_11
Mendeley helps you to discover research relevant for your work.