Paper and invited session @ virtual CDC 2020
- activities
- conferences
- publications
Nils Schlüter presented our paper on "Encrypted explicit MPC based on two-party computation and convex controller decomposition" in an invited session on "Encrypted control and optimization" organized by our group and researchers from the University of Pennsilvania and the Seoul National University.
The paper indroduces a novel encrypted implementation of a model predictice controller (MPC). Roughly speaking, the proposed implementation builds on four observations. First, linear MPC leads to a piecewise affine (PWA) control law. Second, any PWA function can be decomposed into two convex PWA functions (as illustrated in the figure above). Third, each convex part can be expressed as the maximum of finitiely many affine segments. Fourth, twe two resulting maximizations can be securely evaluated using two-party computation. Contact Nils for further details or have a look at the paper:
N. Schlüter and M. Schulze Darup, Encrypted explicit MPC based on two-party computation and convex controller decomposition, in Proc. of the 59th IEEE Conference on Decision and Control (CDC), pp. 5469-5476, 2020. DOI: 10.1109/CDC42340.2020.9304078