A novel clustering scheme for reduced-order macromodeling of massively coupled interconnect structures

An efficient clustering scheme for macromodeling of massively coupled interconnect structures is presented. The proposed method addresses the issue of dense reduced models through a novel reduction algorithm that leads to sparse block-diagonal system matrices. It also overcomes the accuracy degradation, normally associated with the existing port-reduction techniques.