In tone reservation (TR) methods, the high peak-toaverage power ratio (PAPR) of orthogonal frequency division multiplexing (OFDM) signals is reduced through exploiting the reserved tones, i.e., peak-reduction tones (PRT), so as to manipulate the clipping noise generated from the soft limiter with predefined clipping threshold. In this work, from the leastsquares perspective, the PRT are designed with linear operation only and a corresponding clipping threshold is proposed by a simple closed-form formula. Furthermore, a PRT-aided detection algorithm for the receiver is proposed to exploit the information embedded in the PRT, which is a linear function of the datatone signal. The data-tones and PRT are jointly and iteratively estimated to increase the accuracy of detection. No extra receiveside information is required in the proposed method. Simulation results reveal that a remarkable gain in bit error rate (BER) is achieved.