System identification: from blind to informed paradigm

System identification (SI) is necessary in many applications to understand and control the behavior of the considered system. In particular, in the inverse problems it is required to identify the relation between the output and input signals of the system in order to restore or extract some information on the latter.

In many situations, one has to handle the identification problem using only the system output signal in addition to structural or statistical information about the system and its inputs. This is referred to as the blind system identification (BSI) problem.

The BSI problem has been deeply investigated during the last three decades and a plethora of solutions exist in the literature, especially for the parametric linear system case. However, even though this is a mature and well mastered research problem, one can notice that it remains limited in terms of genuine applications due to several reasons including the relatively high computational cost of the blind methods and the poor quality of the identification results in many of the considered real-life situations. Concerning the cost, one can expect reasonably that, thanks to the fast increase of the computational power and the new tools relative to the distributed or parallel computing, this limitation will be less and less restrictive in the future. For the identification quality, the limitation is inherent to the blind processing itself and one has almost reached its affordable performance bounds.

Our objective is to move from the ‘blind processing’ paradigm to the ‘informed’ one by taking advantage of the advances in the learning processes and the fast developing computational facilities.

Selected Publications

Abdulmajid Lawal, Karim Abed-Meraim, Azzedine Zerguine, and Nguyen Linh Trung. Low Cost Blind and Semi-Blind Equalizers for Nonlinear SIMO Systems. IEEE Transactions on Wireless Communications, January 2024 [under review].

Ouahbi Rekik, Kabiru Nasiru Aliyu, Bui Minh Tuan, Karim Abed-Meraim, and Nguyen Linh Trung. Fast subspace-based blind and semi-blind channel estimation for MIMO-OFDM systems. IEEE Transactions on Wireless Communications, December 2023 [under minor revision].

Do Hai Son, Karim Abed-Meraim, Tran Trong Duy, Nguyen Linh Trung, and Tran Thi Thuy Quynh. On the semi-blind mutually referenced equalizers for MIMO systems. In 2023 Asia Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA-ASC), Taipei, Taiwan, November 2023.

Tran Trong Duy, Nguyen Van Ly, Nguyen Linh Trung, and Karim Abed-Meraim. Fisher information estimation using neural networks. REV Journal on Electronics and Communications, vol. 13, no. 1-2, pp.1-10, January-June 2023.

Tran Trong Duy, Nguyen Van Ly, Nguyen Viet Dung, Nguyen Linh Trung, and Karim Abed-Meraim. Fisher information neural estimation. In 30th European Signal Processing Conference (EUSIPCO), Belgrade, Serbia, pp. 2111–2115, August 2022.

Ouahbi Rekik, Karim Abed-Meraim, Mohamed Nait-Meziane, Anissa Mokraoui, and Nguyen Linh Trung. Maximum likelihood based identification for nonlinear multichannel communications systems. Signal Processing, vol. 189, no. 108297, December 2021.

Le Trung Thanh, Karim Abed-Meraim, and Nguyen Linh Trung. Performance lower bounds of blind system identification techniques in the presence of channel order estimation error. In 29th European Signal Processing Conference (EUSIPCO), Durbin, Ireland, pp. 1646-1650, August 2021.

Le Trung Thanh, Karim Abed-Meraim, and Nguyen Linh Trung. Misspecified Cramer–Rao Bounds for Blind Channel Estimation Under Channel Order Misspecification. IEEE Transactions on Signal Processing, vol. 69, pp. 5372-5385, 2021.

Mohamed Nait-Meziane, Karim Abed-Meraim, Zhipeng Zhao, and Nguyen Linh Trung. On the Gaussian Cramér-Rao bound for blind single-input multiple-output system identification: Fast and asymptotic computations. IEEE Access, 8:166503–166512, September 2020.

Other information

Co-PI: Prof. Karim Abed-Meraim