A novel algorithm of inter-subchannel interference self-cancellation for OFDM systems

Time-varying multipath channels distort the orthogonality between subchannels in orthogonal frequency-division multiplexing (OFDM) transmission. The loss of subchannel orthogonality causes inter-subchannel interference (ICI), which limits the achievable bit-error probability (BEP) at high signal-to-noise ratio (SNR). In this paper, we propose a simple but very effective ICI self-cancellation algorithm. A pre-processor and a post-processor are inserted in the transmitter and receiver, respectively. The pre-processor adds diversity to the frequency-domain symbols by time-domain periodical extension, while the post-processor uses this diversity to make most of the ICI self-cancelled. Our algorithm can provide a trade-off between ICI reduction and system throughput by adjusting the length of periodical extension. For the full-extension scheme, we show that the ICI can be completely removed if the channel variation of each path is linear with time within one extended symbol interval. We further propose an equivalent implementation of the post-processor such that the complexity of the receiver is the same as the standard OFDM receiver. This implementation also enables our algorithm to be readily combined with other OFDM algorithms of channel estimation, synchronization, coding, and so on that do not consider the ICI effect. Applying the proposed algorithm of ICI reduction makes these algorithms more applicable in fast-fading channels. To provide more insight on the ICI cancellation, we derive the equivalent channel effect of our algorithm. We also analyze the variance of ICI and observe the density function of the residual ICI in our algorithm, based on which we show a procedure to derive a BEP upper bound. The proposed algorithm is further validated by simulation and the comparison with another ICI self-cancellation algorithm.