Comparison of low complexity clipping algorithms for OFDM

This paper explores low complexity clipping of an orthogonal frequency division multiplexing (OFDM) transmit signal for critically sampled or oversampled complex data as a solution for the peak to average power reduction (PAPR) problem. Four existing clipping algorithms are compared to two new low complexity algorithms in an OFDM environment. Their performance is examined through mathematical analysis and simulation to find the effect of clipping on the signal to noise ratio (SNR). The complexity of each algorithm is compared in terms of hardware operations. One solution is a variation of a Lucent patent called vector subtraction that avoids the division in the scaling operation for a minor reduction in the SNR (<0.5 dB for 3 iterations). The other solution called sector clipping limits the signal to a predefined level without the need of a magnitude estimate, avoiding the need for multipliers, divisions, square roots, or look up tables (LUT). This is at the expense of a 1 to 4 dB (dependant on clip level and number of sectors) decrease in the SNR compared to the optimum clipping approach. Both methods are suitable for low power integrated circuit implementation with low memory requirements and low latency.