Indoor visible light communication has been widely studied for its simultaneous illumination and secure communication functions. In practical indoor multiple-input and multiple-output (MIMO) visible light communication (VLC) environments, users usually concentrate their work in specific areas, resulting in a non-uniform user distribution. As the number of users increases, the way users tend to access to the nearest access point (AP) may overload some APs, a problem that has rarely been considered in most studies. In this paper, we introduce a backtracking AP assignment scheme using a channel gain weighting model, which aims to balance the distribution of connections between APs and users, thereby reducing the load on APs and increasing the system sum rate. However, the power allocation method using power allocation coefficients designed based on user channel gains cannot meet the communication needs of all users. At the same time, users are subject to interference from other users' signals during the communication process, which may affect the rate that users can achieve, depending on the amount of power allocated to other users' signals. Therefore, the scheme of power allocation by the dimension-by-dimension dynamic cosine algorithm (DDSCA) improved by the optimal parameters r1 combined with the optimal solution direction adaptive exploration strategy is adopted. It redistributes the user's power to solve the problem that each user's achievable rate is lower than the threshold value, which can improve the overall communication performance.

The non-uniform distribution of indoor users in MIMO scenarios leads to poor user communication quality due to accessing too many users at APs. To this end, we propose a joint AP assignment and power allocation scheme to improve the sum rate of data transmission in the system. A candidate list ℒ is constructed by generating the channel gain weight parameters of users and APs, and all APs are divided into subsets. The performance of the scheme is dynamically explored based on ℒ for users joining each AP subset. Power allocation coefficients are designed during the dynamic exploration of the AP allocation scheme, and the APs allocate the power to the users based on the power allocation coefficients. Then the AP allocation algorithm based on the channel gain weight model backtracking is adopted to balance the connection distribution between APs and users, and the centralized controller controls the APs to provide communication services for different users. In addition, after obtaining the optimal AP allocation scheme, the DDSCA is improved by designing the transformation parameters and the strategy of adaptive exploration in the direction of the optimal solution to satisfy the communication needs of all users.

In the case of non-uniform distribution of indoor users (Fig. 5), the proposed scheme is compared with other schemes under different assumptions. The results show that the proposed scheme outperforms the other schemes in increasing the system sum rate with a larger number of users (Fig. 6). By introducing an improved DDSCA (IDDSCA) for power allocation, it can effectively improve user satisfaction index. In addition, IDDSCA enables all users to reach the achievable rate threshold (Fig. 9), which leads to an increase in the average user satisfaction index. The system sum rate (Fig. 7) and average user satisfaction indices (Fig. 10) are examined for users equipped with receivers having different field-of-view (FOV) angles. It is observed that a larger FOV angle can have a significant impact on the channel gain, resulting in a decrease in the system sum rate. However, the proposed solution effectively mitigates the impact of FOV on the system sum rate and user satisfaction index through dynamic APs and power allocation. In addition, we analyze the effect of the maximum transmit power of the AP on the system performance (Fig. 8), and as the AP transmit power increases from 6 W to 20 W, the proposed scheme increases the sum rate of the system by 15.62%, which shows that the higher transmit power can allow users to ensure stable communication while bringing a higher achievable rate.

In this paper, we study the problem of maximizing the system sum rate and improving the user quality of service in MIMO VLC systems with non-uniformly distributed indoor users. We propose an AP allocation algorithm based on the backtracking of the channel gain weight model so that the user can select the AP that can improve the system sum rate. To improve the system sum rate while meeting the user's communication requirements, it is necessary to allocate the power to the user after the user accesses to the AP. We propose an IDDSCA, which allocates the power to the user to reach the achievable rate threshold. The analysis examines the system sum rate and user satisfaction of the proposed scheme under various environmental assumptions. Simulation results indicate that the improvement in the proposed AP allocation scheme becomes more pronounced as additional users are added, and the improved power allocation scheme can effectively increase user satisfaction.