A Framework of Estimating Connectivity of Multichannel Data Using MVAR-based Model of Distributed Adaptive Networks

Recognition of brain function and determining its connectivity pattern is one of the most challenging areas in brain signal processing. Researchers have conducted numerous studies in this field, identifying various measures. In this paper, we introduce a novel framework for estimating the connectivity of multichannel data, such as EEG, using distributed adaptive networks based on the MVAR model. The approach views the brain as a dynamic, distributed adaptive network where different nodes collaborate. Therefore, the data of other nodes contributes with specific coefficients in generating the data of each network node. In the proposed model, we used the sparsity to estimate the coefficients. This feature makes the coefficients of connectivity more accurately estimated. Also, we have defined a connectivity measure according to the estimated coefficient matrix. This measure, which considers both the strength and direction of the coefficients, provides a clear procedure for interpreting the connectivity of the multichannel data. The simulation results show that the proposed method provides good performance in different scenarios and is a suitable criterion for evaluating connectivity.