Data-Driven Sensor Selection Method Based on Proximal Optimization for High-Dimensional Data With Correlated Measurement Noise

The present paper proposes a data-driven sensor selection method for high-dimensional nondynamical system with strongly correlated measurement noise. proposed is based on proximal optimization and determines locations by minimizing the trace of inverse Fisher information matrix under block-sparsity hard constraint. can avoid difficulty noise, in which possible must be known advance calculating precision selecting locations. problem efficiently solved alternating direction multipliers, computational complexity proportional to number potential when it used combination low-rank expression noise model. advantage over existing methods demonstrated through experiments using artificial real datasets.