Two of the most important criteria for a sensor array beam-pattern are beamwidth and side-lobe level. A narrower beamwidth means a better angular resolution and well separation of close sources. The beamwidth of a sensor array is directly proportional to array aperture, the beamwidth gets narrower as array aperture increases. However, increase in array aperture means increase in number of used sensors due to Nyquist spatial rate and this means increase in cost and computational load. Therefore, sparse arrays which violate Nyquist spatial rate are used. To cancel grating lobes occurred due to violation of Nyquist rate a weight optimization is implemented. In this paper, an approach which does not need weight optimization for grating lobe cancellation is presented. This method is inspired from nonlinear apodization (windowing) method. The necessary information to cancel grating lobes is obtained from a sub-array contained in the sparse array and this sub-array obeys Nyquist spatial rate.