Stable Dynamic Quantizer Design for MIMO Non-Minimum Phase Systems based on Serial System Decomposition

The optimal dynamic quantizer, supposed to be combined with conventional feedback controller, is a model-based converter from a continuous-valued control input to a discrete-valued one that minimizes the resulting quantization error. A known technical difficulty was the case of non-minimum phase systems that it tends to result in unstable dynamic quantizer due to its model-based nature. In this paper, we propose a method to design a stable dynamic quantizer for multi-input, multi-output (MIMO) non-minimum phase systems. The key technique is an appropriate serial system decomposition, likewise the so-called inner-outer factorization strategy, to separate the system into the minimum phase factor and the non-minimum phase one. We discuss some theoretical issues concerning this approach and show that the optimal dynamic quantizer is only dependent on the outer part.