Counting the active node numbers is an essential parametric quantity as it is used for appraising the performance of a pragmatic network. Different conventional protocol based techniques can be used to determine this parametric quantity. However, these techniques have several limitations including protocol complexity in underwater environment due to the unique characteristics of underwater network such as long propagation delay, high error rate, non-negligible captures effect, dispersion, node mobility and high path loss. In order to evade those limitations a cross-correlation based node estimation approach is implicated. It is an alternative and a statistical approach for signal processing and it is also an efficient technique for any environment networks. In this piece of work, a node estimation technique is proposed by cross-correlating the received signals by four sensors for underwater networks. Nodes and sensors are considered as transmitters and receivers respectively. The aim of this work is to improve the estimation performance including accuracy compared to the performances of existing methods. Here, the performance is measured with respect to the comparison of coefficient of variance and error bar and it exhibits better performance comparing to the existing works.