Block Precoding for Peak-Limited MISO Broadcast VLC: Constellation-Optimal Structure and Addition-Unique Designs

In this paper, we investigate the design of energy-efficient space-time modulation for peak-limited MISO broadcast visible light communication (VLC) systems by cooperatively managing the non-negative multiuser interference. We first characterize a constellation-optimal structure that maximizes the worst-case minimum Euclidean distance of all users for a general space-time modulation design. It turns out that the optimal space-time constellation can be constructed via the spatial repetition of the optimal multidimensional constellation in time dimension over ideal additive white Gaussian noise channels. Then, based on this structure, we specifically design two classes of energy-efficient time-dimensional constellations: 1) for the integer overall bit rate, we design the optimal linear precoded block design, which admits fast maximum likelihood demodulation algorithms. 2) for the non-integer case, we propose a nonlinear precoding scheme called block coded modulation, which sums the code word sets of the optimal linear design and a block channel code. In addition, we show that these two classes of designs are addition-unique, thus, generating an energy-efficient mapping from users’ data to the transmitted constellations. This property enables the efficient demodulation of the sum signal from a noisy received signal as well as the decoding of individual signal from the estimated sum signal. Extensive computer simulations indicate that our addition-unique designs have remarkable performance gains over the currently available zero-forcing, minimum mean square error and time-division multiple access methods for the multiuser multi-input-single-output VLC broadcast systems.