Interference Alignment at Intermediate SNR with Perfect or Noisy CSI

Interference alignment is a new technique combining transmitter precoding and receiver interference suppression to achieve the optimal multiplexing gain in interference networks by exploiting knowledge of channel state information of all transmission links. So far closed form solutions for the transmit filters have only been found in certain cases. Also the feasibility of interference alignment schemes based on symbol extensions, over a limited number of signalling dimensions, is still an open problem. In this work we investigate the performance in terms of bit error rates, of interference alignment schemes at intermediate signal-to-noise ratios, through Monte Carlo simulations. We focus our attention on the three and four users time-varying interference channel, using both the closed form solutions known at present as well as iterative algorithms. We then investigate the impact of noisy channel state information on the performances of some of the interference alignment systems considered. In the single input single output interference channel the closed form solutions of the interference alignment cause considerably different bit error rates for the different nodes in the network. In the multiple input multiple output interference channel we exhibit that bit error rate saturates at moderate signal-to-noise ratios when interference alignment schemes are infeasible and even when they are feasible, some of the analyzed algorithms show unpredictable behaviors by deteriorating the performance as the signal-to-noise ratio exceeds a threshold. Further refinements are necessary in order to obtain better bit error rates in these cases. We evince that additional improvements are also needed to the original interference alignment schemes in order to mitigate their sensitivity to noisy channel state information.