Prior-Knowledge Incorporated Forward and Reverse Neural Network Modeling for Estimating Performance Parameters of Band-Notched UWB MIMO Antennas: Smart estimation strategies

The design and analysis of multiple-input, multiple-output (MIMO) antennas using an electromagnetic (EM) simulator is a computationally expensive process. In this case, an artificial neural network (ANN) can be effectively used to obtain a surrogate model. The accuracy and requirement of large data samples can be improved by incorporating prior knowledge into the ANN model. Hence, in this article, a prior knowledge input (PKI) method-based ANN model has been presented for modeling MIMO antennas. Two ultrawideband (UWB) MIMO antenna designs with band-notch characteristics have been proposed. The MIMO antenna designed using the FR4 substrate has four ports and rejects two interfering bands: X-band downlink (6.3–7.6 GHz) and uplink (8.05–8.7 GHz). The analysis and synthesis of the MIMO antennas are achieved using the PKI forward and reverse modeling methods. The performance of the PKI model is found to outperform the standard ANN model in terms of the error variation. Also, prototypes of both the MIMO antennas are fabricated and characterized. The simulated, computed, and measured counterparts are found to be in good agreement with each other.