The capacity of fast fading channels using lattice codes: Is separability necessary?

This paper considers communication over the point-to-point Gaussian channel with stationary and ergodic channel gain and full channel state information. The capacity of this chan­nel can be achieved in a straightforward manner via any capacity-achieving codebook for the non-fading Gaussian channel, in conjunction with separable coding, i.e., coding independently over each fading state. Despite its elegance, separable coding causes substantial delays in the communication process. In this paper, we propose a lattice coding scheme that achieves the fast fading channel capacity without separable coding. Interestingly, the decision regions used are universal for a given channel distribution, as long as the sequence of channel realizations is robustly typical. Some of the techniques used in the achievability proof are of independent interest and can be applied to a variety of channels as well as different classes of codebooks.

[1]  Aria Nosratinia,et al.  Approaching the ergodic capacity with lattice coding , 2014, 2014 IEEE Global Communications Conference.

[2]  Hans-Andrea Loeliger,et al.  Averaging bounds for lattices and linear codes , 1997, IEEE Trans. Inf. Theory.

[3]  Giuseppe Caire,et al.  Lattice coding and decoding achieve the optimal diversity-multiplexing tradeoff of MIMO channels , 2004, IEEE Transactions on Information Theory.

[4]  Yuval Kochman,et al.  Lattice Coding for Signals and Networks: Side-information problems , 2014 .

[5]  Sergio VerdÂ,et al.  Fading Channels: InformationTheoretic and Communications Aspects , 2000 .

[6]  Pravin Varaiya,et al.  Capacity of fading channels with channel side information , 1997, IEEE Trans. Inf. Theory.

[7]  Wayne E. Stark,et al.  Channels with block interference , 1984, IEEE Trans. Inf. Theory.

[8]  Alon Orlitsky,et al.  Coding for computing , 1995, Proceedings of IEEE 36th Annual Foundations of Computer Science.

[9]  Simon Litsyn,et al.  Lattices which are good for (almost) everything , 2005, IEEE Transactions on Information Theory.