In this paper, the performance of a multicarrier underlay cognitive radio system is analyzed considering that the successive interference cancellation (SIC) method is used for detection. In the proposed system model, a multicarrier-based secondary user (SU) uses all the subcarriers of the primary user (PU) in underlay mode, guaranteeing the quality of service (QoS) to the primary system. The power allocation strategy is modified accordingly. With a fixed PU transmit power, power is allocated to the SU subcarriers to maximize its throughput while ensuring minimum throughput to PU. The system is analyzed for its performance considering an L-tap multipath Rayleigh fading channel. Analytical expression for outage probability at SU is derived in closed-form. Also, analytical expressions and closed-form approximations for average symbol error rates (SER) per subcarrier of PU and SU are obtained. The derived expressions are validated with the Monte-Carlo (MC) simulations. Further, the proposed scheme was implemented in an orthogonal frequency division multiplexing (OFDM)-based system with different detection schemes used at the receiver. SERs were compared through extensive simulations and it was noticed that SIC with channel estimation based on minimum-mean-squared-error (MMSE) often performed better than that based on least squares (LS) estimation. Observations reveal that the nominal error rates were achieved on the optimal choice of the parameters. The proposed system ensures sufficient throughput to SU, meanwhile eliminating the need for opportunistic sensing.INDEX TERMS NOMA, outage probability, power allocation, SER, underlay cognitive radio.