Layered Space-Time Index Coding

TL;DR

This paper introduces layered space-time index coding (LSTIC), a novel framework for MIMO multicasting that leverages side information to exponentially improve code performance, preserving properties of perfect space-time block codes.

Contribution

It proposes a general LSTIC construction from various space-time block codes, analyzing its exponential side information gain and demonstrating its effectiveness over well-known codes.

Findings

LSTIC provides exponential determinant gains with side information.

LSTIC over perfect STBCs retains perfect code properties.

Simulation confirms predicted side information gains.

Abstract

Multicasting $K$ independent messages via multiple-input multiple-output (MIMO) channels to multiple users where each user already has a subset of messages as side information is studied. A general framework of constructing layered space-time index coding (LSTIC) from a large class of space-time block codes (STBC), including perfect STBC, is proposed. We analyze the proposed LSTIC and show that it provides minimum determinant gains that are exponential with the amount of information contained in the side information for any possible side information. When constructed over a perfect STBC, the proposed LSTIC is itself a perfect STBC and hence many desired properties are preserved. To illustrate, we construct LSTIC over the following well-known STBCs: Golden code; $3\times 3$, $4\times 4$, $6\times 6$ perfect STBCs; and Alamouti code. Simulation results show that the obtained side information gain can be well predicted by our analysis.