In cognitive radio, current trend is to utilize geolocation database for TV bands. Considering more dynamic bands in terms of primary user activity, however, necessitates the use of Radio Environment Map (REM), which is an advanced knowledge base that stores live multidomain information on the entities in the network and the environment. In Cognitive Radio Networks (CRNs), mobile nodes that are capable of measuring the energy of the frequency bands are less capable compared to dedicated sensing nodes in the network. Therefore, deployment algorithm of the dedicated sensor nodes is of great importance and affects the constructed REM interference map quality. We propose a novel deployment algorithm for CRNs that considers user distribution probabilities. Numerical results confirm that the proposed deployment algorithm significantly improves the REM performance. The proposed algorithm is compared with random deployments and it is applied on Kriging and LIvE REM construction techniques.