Optimal input design for worst-case system identification in \(l_ 1/l_ 2/l_ \infty\) (Q2367117)

The problem of optimum experimental design for worst-case SISO finite impulse response system identification is considered, with the aim of minimizing the worst identification error under deterministic error assumptions. The optimal input sequences, minimizing the worst-estimate error in \(l_ 1\), \(l_ 2\) and \(l_ \infty\) norm under bounded (in \(l_ \infty\)) measurement error are investigated, and the minimum number of measurements ensuring the achievement of minimal error is derived. The structure of proper input designs is established.