Determinant-Based Fast Greedy Sensor Selection Algorithm

In this paper, the sparse sensor placement problem for least-squares estimation is considered, and previous novel approach of selection algorithm extended. The maximization determinant matrix which appears in pseudo-inverse operations employed as an objective function present extended approach. procedure corresponding proved to be mathematically same that previously proposed QR method when number sensors less than state variables (undersampling). On other hand, authors have developed a new greater (oversampling). Then, unified formulation two algorithms derived, lower bound given by shown using monotone submodularity function. effectiveness on real datasets demonstrated comparing with results algorithms. numerical show improves error approximately 10% compared conventional methods oversampling case, where defined ratio difference between reconstructed data full observation observation. For NOAA-SST problem, has more ten thousand candidate points, selects positions few seconds, required several hours case 3.40 GHz computer.