Research On Optimal Power Allocation Scheme For Indoor NOMA-VLC Systems

In the face of the future wireless networks' service requirements,including massive connectivity,enhanced mobile broadband,ultra-reliability,low-latency and so on,visible light communication(VLC)and non-orthogonal multiple access(NOMA)have emerged as a promising communication paradigm and an outstanding multiple access scheme,respectively.VLC can implement communication and illumination concurrently,and NOMA can facilitate both the spectral efficiency and the user experience.NOMA can be ideally suited to indoor VLC systems due to limited user number in each optical attocell,high signalto-noise ratio of the VLC link,tunable LED's transmission angle and receiver's field of view,and quasi-static channel state information.Therefore,from perspective of enhancing the spectral efficiency and exploring the new spectrums,focusing on NOMA-based indoor VLC systems can efficiently resolve the current conflict between data rate requirements and spectrum scarcity.As a key issue in indoor NOMA-VLC systems,power allocation scheme has a significant influence on system sum rate,energy efficiency,bit-error-ratio,outage probability,and user fairness.In addition,NOMA has the dominant advantages over the conventional multiple access schemes in system sum rate and user fairness.This dissertation mainly focused on the optimization of power allocation aiming at achieving the balance between system sum rate and user fairness.Specifically,the following innovative contributions were made.(1)A reasonable channel model is the basis of all theoretical research.However,the traditional indoor VLC channel models generally did not consider the shadowing caused by the randomly mobile multi-users,which is just the common configuration in NOMA systems.This issue was addressed in the dissertation.Firstly,in order to characterize the size and position of randomly mobile multi-users,the probability density functions were introduced to describe the joint distribution of a user's width and height,and the joint distribution of a user's horizontal axis and vertical axis.All users were assumed to be independent identically distributed.Secondly,based on the principle of geometry and mathematic statistics,the weighting function of each link was derived to describe the shadowing caused by the randomly mobile multi-users.Furthermore,the weighting function was equal to the probability of the link being not blocked by any user.Finally,by calculating the weighted sum of all links in the traditional indoor VLC channel model,a novel indoor VLC channel model with shadowing by mobile multi-users was proposed.The expressions of the direct current channel gain and the impulse response were also given.Moreover,the validity of the proposed model was verified through numerical simulation.As was shown,the existence of randomly mobile multi-users indeed led to obstruction of some optical links and degraded the channel quality.(2)In prior studies,fairness in NOMA systems was considered from the perspective of proportional fairness,max-min fairness,maximizing sum logarithmic utility function of user rate,or guaranteeing that all the multiplexed users can achieve a capacity at least as good as orthogonal multiple access.Different from those works,the explicit and determinate mathematical comprehensive assessment of system sum rate and user fairness was implemented in this dissertation to facilitate the power allocation.Firstly,in order to guarantee the numeric scope of normalized user rates,user fairness was described as the gap between the minimum rate and the maximum rate among all users.Secondly,the power allocation to balance system sum rate and user fairness was modeled as an issue of multiattribute decision making(MADM).In the MADM issue,system sum rate and user fairness were the decision parameters,and the sets of power allocation factors were the candidates.Finally,the above MADM issue was solved by introducing the standard deviation method and the technique for order preference by similarity to ideal solution(TOPSIS),and the MADM-based optimal power allocation algorithm was proposed.Specifically,the author enumerated the sets of discrete power allocation factors as the candidates,and calculated the corresponding system sum rate and user fairness for each candidate.Based on the calculated data set,the objective weights of system sum rate and user fairness were derived by using the standard deviation method.Then,the optimal candidate,i.e.,the optimal power allocation scheme,could be found by means of TOPSIS.Through numerical simulation,it could be observed that the MADM-based power allocation algorithm can greatly improve user fairness performance at a small cost of system sum rate.That is to say,a better balance between them could be achieved.As the user number increased,the performance gain grew rapidly,and even the simultaneous gain in system sum rate and user fairness could be achieved.(3)Due to the enumeration of discrete power allocation factors,the above MADM-based power allocation algorithm would face a compromise between the level of performance improvement and computation complexity.A further study was conducted to overcome the above drawback and take the practical application demand into account that user fairness should be higher than a threshold.In addition,the quality of service(QoS)requirements of all users were also considered.Firstly,one optimization problem of user fairness-guaranteed sum rate maximization was formulated,wherein the threshold guaranteeing user fairness could be flexibly adjusted according to the actual requirements.Secondly,for the cases that the threshold took certain special values,the closed-form expression or the solving procedure of the optimal solution was obtained through strict mathematical derivation and optimization analysis.Finally,for the cases that the threshold took certain general values,the differential evolution algorithm was involved,and wherein the operations including variable transformation,population initialization,individual selection in each iteration,and optimal solution selection were improved adaptively.After all,the first differential evolution-based optimal power allocation algorithm was proposed.As numerical simulation results showed,under user fairness constraint of no less than that achieved by conventional algorithms or the optimal power control(OPC)algorithm,the proposed algorithm could always achieve better sum rate performance.The performance gain increased with increasing user number.In addition,the proposed algorithm could also increase the system coverage probability efficiently for application scenarios with guaranteed QoS.(4)Administrators' subjective preference between system sum rate and user fairness was considered,and more practical application requirements were further accommodated.For application scenarios where system sum rate should be higher than a threshold,the other optimization problem of sum rate-guaranteed user fairness maximization was formulated.The threshold guaranteeing system sum rate could be flexibly adjusted according to the actual requirements.Similarly,through the classified discussion around the threshold,adaptive mathematical derivation,optimization analysis and differential evolution algorithm were given.As a result,the second differential evolution-based optimal power allocation algorithm was proposed.As numerical simulation results showed,under sum rate constraint of no less than that achieved by conventional algorithms or the OPC algorithm,the proposed algorithm could always achieve better user fairness performance.The performance gain increased with increasing user number.In the same way,the proposed algorithm could also increase the system coverage probability efficiently for application scenarios with guaranteed QoS.