The paper proposes a scheme to minimize the co-tier and cross-tier interference by properly allocating the sub-channels and power among the femtocell users (FUs) in an OFDMA based cognitive femtocell network. An efficient graph coloring based sub-channel allocation scheme is adapted in which the sub-channels in the uplink band of the macrocell are shared by the femtocells dynamically in an opportunistic way through spectrum sensing. A price-based power allocation scheme using game theory is proposed to assign the transmission power among the FUs, whereby the macrocell base-station (MBS) controls the transmission power of the FUs by pricing their resulted interference power levels at the MBS with the constraint that the total interference created at the MBS in each sub-channel is kept below a tolerable threshold. The game theoretic approach ensures that no FUs can improve its utility by changing its own transmission power selfishly. The allocation of sub-channels among the FUs is centrally controlled whereas the power allocation is handled in a distributed way which makes the process efficient. Finally, numerical examples are presented to analyze the proposed scheme. The results show that substantial number of FUs with good quality of services can be accommodated in the macrocell reusing the uplink frequency sub-channels and thereby enhancing the throughput and coverage of the network.