Research On Optimal Constellation For Multi-Input Multi-Output Visible Light Communication Systems

Visible light communication(VLC),as a novel wireless communication technique,has a promise to become the complement of the present radio frequency(RF)communication system owing to its potential ability to offer ultra-high data rates.However,since most VLC systems utilize intensity modulation and direct detection(IM/DD),channels in multi-input multi-output(MIMO)VLC systems are highly correlated when compared to those in RF-MIMO ones,which bring severe impacts to the performance of conventional MIMO techniques deployed in this field.In order to overcome this problem,in this paper,the static line of sight(LOS)characteristics of the indoor VLC link is exploited and optimal constellations designed for VLC-MIMO systems under different channel states are proposed,details of which are given as follows:1.Owing to the maximum likelihood receiver employed in the visible light communication system,the performance of it is greatly determined by the minimum Euclidean distance among different received signal points.Therefore,the goal of this paper is to maximize this value under a total optical emission power constraint.However,since the corresponding optimal problem belongs to the classic nonconvex quadratically constrained quadratic programming problems,it is generally difficult to be solved.Hence,a Monte Carlo random search(MCRS)algorithm combining with the convex-concave procedure(CCP)is proposed.Compared to other three schemes(repetition coding,spatial modulation and spatial multiplexing),simulation results showed that constellations designed by this algorithm perform the best in all the given cases.2.An indoor real-time 2×2 VLC-MIMO system with constellation size being 2 is considered,of which only one Euclidean distance needs to be optimized.Then,as each entry of the channel matrix is nonnegative and real,some parameters describing the characteristics of the channel can be obtained by eigenvalue decomposition.Using the simple variable transformation,the original optimal problem can be transferred as different maximum problems with different values of those parameters,and the best result can be easily attained.Compared to other two conventional VLC-MIMO techniques(repetition coding and space-shift keying modulation),simulation results showed its best performance in both static and dynamic channels.3.An indoor real-time 2×2 VLC-MIMO system with constellation size being 4 is considered,where six candidates of the minimum Euclidean distance need to be taken into account.In this case,to solve the problem directly is difficulty,and by using MCRS-CCP will take too much time.Therefore,a rapid algorithm is sought in this paper.Firstly,some characteristics of optimal signal constellations under several typical channel states are exploited by deploying MCRS-CCP.Then,based on those findings,an adaptive algorithm on rapid constellation selecting(ARCS),containing merely four candidates,is proposed.Last,simulation results verified its effectiveness in both two channel states.