Characterization and Parameter Estimation of Multiple Input Multiple Output (MIMO) Systems

摘要

Model representations and estimation methods for multiple input multiple output (MIMO) systems are summarized. When extending the identification/estimation procedures from single to MIMO-processes, problems grow considerably. For model selection this is particularly due to the number of parameters needed for a model structure that leaves open all feasible (but probably nonexistent) interconnections, interactions, and noise-influences. This implies the desire to include as much a priori knowledge as possible in the selection of the model structure. For parameter estimation, it is desirable to choose the model as being linear-in-the-parameters, in order to attain simple (recursive) estimation algorithms. Unfortunately, this is not compatible with the requirements with respect to the restriction of the number of parameters. For example, a Markov parameter model has the linearity property, but it has a large (infinite) number of parameters.