Gaussian Interference Channel Capacity to Within One Bit

• Published in: IEEE transactions on information theory Vol. 54; no. 12; pp. 5534 - 5562
• Main Authors: Etkin, R.H., Tse, D.N.C., Hua Wang
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.12.2008; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Additive white noise; Applied sciences; Asymptotic properties; AWGN; Capacity region; Channel capacity; Channels; Decoding; Degrees of freedom; Engineering profession; Exact sciences and technology; Frequency; Gaussian; Gaussian interference channel; Gaussian noise; generalized degrees of freedom; Information science; Information theory; Information, signal and communications theory; Interference; Interference channels; Mathematical analysis; Normal distribution; Optimization; Parameter estimation; Radio spectrum management; Signal to noise ratio; Systems, networks and services of telecommunications; Telecommunications; Telecommunications and information theory; Transmission and modulation (techniques and equipments); Gaussian interference channel; generalized degrees of freedom; Channel capacity; Capacity region; Gaussian channel; Signal to noise ratio
• ISSN: 0018-9448; 1557-9654
• DOI: 10.1109/TIT.2008.2006447

The capacity of the two-user Gaussian interference channel has been open for 30 years. The understanding on this problem has been limited. The best known achievable region is due to Han and Kobayashi but its characterization is very complicated. It is also not known how tight the existing outer bounds are. In this work, we show that the existing outer bounds can in fact be arbitrarily loose in some parameter ranges, and by deriving new outer bounds, we show that a very simple and explicit Han-Kobayashi type scheme can achieve to within a single bit per second per hertz (bit/s/Hz) of the capacity for all values of the channel parameters. We also show that the scheme is asymptotically optimal at certain high signal-to-noise ratio (SNR) regimes. Using our results, we provide a natural generalization of the point-to-point classical notion of degrees of freedom to interference-limited scenarios.