The Compute-and-Forward Transform

We derive an achievable rate region for the Gaussian K-user multiple-access channel (MAC) where all users transmit codewords from a chain of nested lattices. For any set of channel coefficients, this rate region contains points within a constant gap from the sum capacity boundary of the MAC. The main tool used is the recently proposed compute-and-forward framework. A new transformation of a MAC to a modulo-lattice multiple-input multiple-output (MIMO) channel is introduced based on this framework. Specifically, from one noisy linear combination of the transmitted signals the receiver attempts to decode K linearly independent equations with integer-valued coefficients. While the individual rates at which these equations can be decoded are highly sensitive to the exact channel gains, their sum is always within a constant gap from the sum capacity boundary of the MAC. The transformation is then utilized for establishing the desired rate region.