Deployment of small cell base stations (BSs), like pico BSs (PBSs), alongside macro BSs (MBSs) in cellular network leads to increased data rates for the users. However, asymmetric transmission power of BSs and received power based user association (UA) necessitates some form of user offloading between MBSs and PBSs on both the downlink and uplink to maximize system performance. In this work, we formulate optimal downlink and uplink decoupled UA as an optimization problem. We show that this network wide optimization problem is not real time solvable. Hence, we propose a novel fuzzy logic based cell selection framework that can be tuned for system level performance gains. Further, we show that in certain scenarios, careful tuning of fuzzy logic system based UA results in significant increase in system performance compared with existing techniques. © 2016 IEEE.