This study proposes a novel network growth model named ComAwareNetGrowth which aims to mimic evolution of real-world social networks. The model works in discrete time. At each timestep, a new link (I) within-community or (II) anywhere in the network is created (a) between existing nodes or (b) between an existing node and a newcoming node, based on (i) random graph model, (ii) preferential attachment model, (iii) a triangle-closing model, or (iv) a quadrangle-closing model. Parameters control the probability of employing a particular mechanism in link creation. Experimental results on Karate and Caltech social networks shows that the model is able to mimic real-word social networks in terms of clustering coefficient, modularity, average path length, diameter, and power law exponent. Further experiments indicate that ComAwareNetGrowth model is able to generate variety of synthetic networks with different statistics.