Identification of ARMA Models using Intermittent and Quantized Output Observations

DAMIÁN MARELLI; KEYOU YOU; MINYUE FU

Praga

Conferencia; IEEE Conference on on Acoustics, Speech, and Signal Processing (ICASSP); 2011

Institute of Electrical and Electronics Engineers (IEEE)

This paper studies system identification of ARMA models whoseoutputs are subject to finite-level quantization and random packetdropouts. A simple adaptive quantizer and the corresponding recur-sive identification algorithm are proposed and shown to be optimalin the sense of asymptotically achieving the minimum mean squareestimation error. The joint effects of finite-level quantization andrandom packet dropouts on identification accuracy are exactly quan-tified. The theoretic results are verified by simulations.