Study On Resource Allocations For Visible Light Communication Networks

The visible light communication(VLC)is an emerging technique to potentially offer larger transmission capacity,and networking is the first step to the practical VLC system.The ultra-dense indoor VLC network provides high throughput per area,while the VLC communication offers high data-rate links for the outdoor vehicular network.The contents of this paper mainly contain the inter-cell interference coordination in in-door VLC network,the channel compression in VLC MIMO system and the distributed resource allocation in outdoor hybrid VLC/Radio Frequency(RF)vehicular network.This thesis address the inter-cell interference coordination for the indoor attocell multi-color visible light communication(VLC)network under lighting constraints.In the multi-color VLC system,soft frequency reuse-based interference coordination is adopted,which adjusts the signal deviations transmitted on cell-edge color and cell-center color.A static scheduler is proposed to increase the signal to interference plus noise ratio(SINR)for the cell-edge region,given that the SINR for the cell-center re-gion is larger than a certain threshold.To adapt to the interference intensity that varies with the density of base stations,we propose a dynamic scheduler for the inter-cell in-terference coordination.The dynamic scheduler comprises of two separate parts:the distributed part that adjusts the power allocation and asks to restrict color usages in neighboring attocells,and the centralized part that resolves the color conflicts and allo-cates the color resources in the network.Two solution approaches are proposed for the centralized part,one based on the linear programming(LP)relaxation of the integer pro-gramming,and the other based on the greedy color allocation.Simulation results show that the dynamic scheduler can improve the cell-edge user's throughput with acceptable penalty to the overall throughput.This thesis address the feedback information reduction and analyze induced rate loss for the indoor multi-color visible light communication(VLC)multi-input multi-output(MIMO)system.The principle component analysis(PCA)is applied to com-press the MIMO channel estimated at the receiver,which allows the feedback overhead to be efficiently reduced.However,due to the information loss in PCA,the compres-sive achievable rate based on recovered MIMO channel at the transmitter is different with the real one,which may leads to the unreliable downlink transmission.Hence we analyze the rate loss induced by PCA channel compression,and ensure that the practi-cal transmission rate,obtained by subtracting the compressive achievable rate with the analyzed rate loss,is smaller than the real rate.The simulation results show that the PCA-based compression method significantly reduces the feedback overhead of mas-sive VLC MIMO system.The practical transmission rate is always smaller than the real transmission rate,which promises the downlink zero-error transmission.This thesis address joint channel allocation and rate control to maximize the net-work throughput in vehicular ad hoc networks(VANET).Considering the limited chan-nel resources and Quality of Service(QoS)requirements of vehicular communication,a mixed-integer linear programming(MILP)problem is formulated.However,it's unre-alistic and inefficient to choose a center node to perform global optimization in VANET.Hence,we propose a distributed channel allocation and rate control approach to solve the cross-layer design problem.Applying the carrier sense multiple access with col-lision avoidance protocol,the proposed channel allocation decides the channel usages by measuring the channel demands of links,and transforms the MILP problem to the corresponding linear programming(LP)problem.We adopt the alternating direction method of multipliers(ADMM)to optimize the flow rates.Since imbalanced rates may be allocated to paths passing the same link,rate adjustment is proposed to ensure the throughput requirement of each path.Simulation results demonstrate the fast conver-gence of the proposed distributed algorithm,and the improvement of ADMM with rate adjustment over the ADMM.While the performance is sensitive to the outage prob-ability of the VLC link,increasing the VLC link rate does not substantially improve the probability of finding a feasible solution.In addition,a small radio frequency(RF)communication range is desirable for a network with dense node distribution due to the dominated interference.