Rate loss caused by limited feedback and channel delay in interference alignment

Interference alignment (IA) promises substantial theoretical gain to achieve the maximum degrees of freedom (DoF) in interference channel. However, perfect global channel state information (CSI) is required, which is difficult to achieve in practical wireless communication system. Imperfect channel knowledge severely degrades the gain of IA, this paper investigates the performance of IA with limited feedback and CSI delay. Given imperfect CSI caused by the limited feedback and channel delay, the expression of data rate is derived. Compared to the rate with perfect CSI, a rate loss upper bound is obtained by analytical derivation. Furthermore, we show that if the time delay and signal-to-noise-ratio (SNR) satisfy a certain relationship, the system rate loss can maintain a constant value, and with SNR increases, optimal transmission DoF can be obtained. Extensive simulations demonstrated that the rate loss reduces when the number of feedback bits increases and the channel delay decreases. Meanwhile, as SNR increases, the system performance with imperfect CSI improves more slowly than with perfect CSI.