Two-User MIMO Broadcast Channel with Transmit Correlation Diversity: Achievable Rate Regions

Khac-Hoang Ngo (1) , Fan Zhang (2) , Sheng Yang (3) , Aria Nosratinia (2)

1 Chalmers University of Technology [Gothenburg, Sweden]
2 UT Dallas - University of Texas at Dallas [Richardson]
3 L2S - Laboratoire des signaux et systèmes

In a multiple-input multiple-output (MIMO) broadcast channel (BC), the difference in spatial transmit correlation matrices of different users is called transmit correlation diversity. Recently, several works have extended this concept beyond its original scope, to include channels whose transmit correlation matrices have non-overlapping eigenspaces. In contrast to earlier analyses of overlapping eigenspaces that were mostly described in terms of degrees-of-freedom, this work presents achievable rate regions. These achievable regions are derived by ratesplitting, product superposition, or a combination thereof. Our rate expressions make explicit the contribution of the common parts and individual (non-overlapping) parts of the correlation eigenspaces toward the achievable rate region. As a by-product, a result of Hassibi and Hochwald on MIMO channel training is extended to channels with spatial correlation.

Domains: Information Theory [math.IT]

Khac-Hoang Ngo, Fan Zhang, Sheng Yang, Aria Nosratinia. Two-User MIMO Broadcast Channel with Transmit Correlation Diversity: Achievable Rate Regions. 2021 IEEE Information Theory Workshop (ITW2021), Oct 2021, Kanazawa (virtual), Japan. pp.1-6, ⟨10.1109/ITW48936.2021.9611406⟩. ⟨hal-03420090⟩