Some new approaches for the implementation of optimum MTI filters

Implementation of optimum MTI filters is rendered difficult due to the non-homogeneous clutter distribution. The sheer magnitude of the computational effort and memory involved for the implementation of the optimum processor makes it highly impractical. The aim of this paper is to obtain the optimum solutions in easily realizable forms. It is shown here that simple implementation of the optimum (or adaptive) solutions can be obtained without recourse to real-time and complex eigenvector computation or even to LMS - type algorithms. The optimum coefficient computations only involve averaging and table-look-up operations. The memory requirement is reduced to the order required for implementation of conventional MTI filters, viz., O(Nr), where Nris the number of range cells. This is against O(NrNb) usually needed for optimum filters where Nbis the number of bearing cells.