Product Superposition for MIMO Broadcast Channels

TL;DR

This paper introduces a novel multiplicative superposition technique for MIMO broadcast channels with unequal CSIR, achieving optimal degrees of freedom and outperforming orthogonal transmission in certain antenna configurations.

Contribution

It proposes a new superposition scheme for MIMO broadcast channels with asymmetric CSIR, demonstrating its optimality and constructing two signaling methods that improve DoF over traditional approaches.

Findings

Superposition achieves optimal DoF in various configurations.

Two signaling schemes outperform orthogonal transmission.

Interference cancellation enhances DoF in the second scheme.

Abstract

This paper considers the multiantenna broadcast channel without transmit-side channel state information (CSIT). For this channel, it has been known that when all receivers have channel state information (CSIR), the degrees of freedom (DoF) cannot be improved beyond what is available via TDMA. The same is true if none of the receivers possess CSIR. This paper shows that an entirely new scenario emerges when receivers have unequal CSIR. In particular, orthogonal transmission is no longer DoF-optimal when one receiver has CSIR and the other does not. A multiplicative superposition is proposed for this scenario and shown to attain the optimal degrees of freedom under a wide set of antenna configurations and coherence lengths. Two signaling schemes are constructed based on the multiplicative superposition. In the first method, the messages of the two receivers are carried in the row and column spaces of a matrix, respectively. This method works better than orthogonal transmission while reception at each receiver is still interference free. The second method uses coherent signaling for the receiver with CSIR, and Grassmannian signaling for the receiver without CSIR. This second method requires interference cancellation at the receiver with CSIR, but achieves higher DoF than the first method.