Rate splitting is approximately optimal for fading Gaussian interference channels

In this paper, we study the 2-user Gaussian interference-channel with feedback and fading links. We show that for a class of fading models, when no channel state information at transmitter (CSIT) is available, the rate-splitting schemes for static interference channel, when extended to the fading case, yield an approximate capacity region characterized to within a constant gap. We also show a constant-gap capacity result for the case without feedback. Our scheme uses rate-splitting based on average interference-to-noise ratio (inr). This scheme is shown to be optimal to within a constant gap if the fading distributions have the quantity log (E [inr]) - E [log (inr)] uniformly bounded over the entire operating regime. We show that this condition holds in particular for Rayleigh fading and Nakagami fading models. The capacity region for the Rayleigh fading case is obtained within a gap of 2.83 bits for the feedback case, and within 1.83 bits for the non-feedback case.

[1]  Mehul Motani,et al.  On The Han–Kobayashi Region for theInterference Channel , 2008, IEEE Transactions on Information Theory.

[2]  Amir Salman Avestimehr,et al.  Capacity Results for Binary Fading Interference Channels With Delayed CSIT , 2013, IEEE Transactions on Information Theory.

[3]  Reza K. Farsani The capacity region of the wireless ergodic fading Interference Channel with partial CSIT to within one bit , 2013, 2013 IEEE International Symposium on Information Theory.

[4]  Myung Gil Kang,et al.  Ergodic Interference Alignment With Delayed Feedback , 2013, IEEE Signal Processing Letters.

[5]  Suhas N. Diggavi,et al.  Bursty interference channel with feedback , 2013, 2013 IEEE International Symposium on Information Theory.

[6]  D. Tuninetti,et al.  Gaussian fading interference channels: Power control , 2008, 2008 42nd Asilomar Conference on Signals, Systems and Computers.

[7]  Sae-Young Chung,et al.  Aligned interference neutralization and the degrees of freedom of the 2 × 2 × 2 interference channel , 2010, 2011 IEEE International Symposium on Information Theory Proceedings.

[8]  David Tse,et al.  Feedback Capacity of the Gaussian Interference Channel to Within 2 Bits , 2010, IEEE Transactions on Information Theory.

[9]  Hua Wang,et al.  Gaussian Interference Channel Capacity to Within One Bit , 2007, IEEE Transactions on Information Theory.

[10]  Necdet Batir,et al.  Inequalities for the gamma function , 2008 .