Affiliation:
1. Tianjin University
2. 54th Research Institute of China Electronics Technology Group Corporation
3. State Grid Chongqing Electric Power Company
Abstract
To enhance the throughput of visible light communication (VLC), this
paper investigates precoding and power allocation jointly in a free
space optical channel with signal-dependent noise (SDN-FSO). With the
presence of SDN from a light source, a more realistic multi-user
multiple-input single-output (MU-MISO) VLC channel model is
introduced. First, based on the entropy power inequality, a capacity
bound for the SDN-FSO channel is derived under a MU-MISO scenario.
Second, due to the nonnegativity of SDN-FSO input, a zero forcing
constraint is introduced to alleviate multi-user interference. Third,
from the perspective of resource allocation, min-max and total power
budget constraints are taken into consideration. A non-convex joint
optimization problem is formulated and divided into two subproblems,
i.e., the precoding subproblem and power allocation subproblem.
Moreover, an iterative algorithm based on a concave and convex program
and a simplified high signal-to-noise ratio approximation method are
iteratively implemented in the precoding subproblem. For the power
allocation subproblem, successive convex approximation is employed to
ensure its optimality. Simulation results indicate that the proposed
joint optimization methods can converge after about two iterations and
accomplish a throughput improvement of up to 1∼3nats/s/Hz as compared with the benchmarks for
four different spacing scenarios. When channel state information is
interfered with noise, the proposed methods still demonstrate their
superior performance with varying degrees of noise.
Funder
State Grid Electric Power Co.,
Ltd
Foundation of Shenzhen
City
54th Research Institute of China
Electronics Technology Group Corporation
National Natural Science Foundation of
China
Subject
Atomic and Molecular Physics, and Optics,Engineering (miscellaneous),Electrical and Electronic Engineering
Cited by
2 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献