This work presents a new array geometry, which is capable of providing $O(M^2N^2)$ degrees of freedom (DOF) using only $MN$ physical sensors via utilizing the second-order statistics of the received data. This new array is composed of multiple, identical minimum redundancy subarrays, whose positions follow a minimum redundancy configuration. Thus the new array is a minimum redundancy array (MRA) of MRA subarrays, and is termed {\em nested MRA}. The sensor positions, aperture length, and the number of DOF of the new array can be predicted if these parameters of MRA subarrays are given. Numerical simulations demonstrate the superiorities of the proposed array geometry in resolving more sources than sensors and DOA estimation.