Objective As a new type of optical wireless communication technology, visible light communication is becoming a popular communication technology due to its lighting characteristics, communication characteristics, and broad spectra. In intelligent transportation, visible light communication can use car lights as signal transmitters for information transmission. Therefore, the application of visible light communication to the internet of vehicles has broad development prospects. Visible light communication has a limited communication range because its communication carrier is visible light. The communication performance is reduced due to the problem of external visible light interference. When using visible light for vehicle communication, it is necessary to supplement it with other networks. In the communication process, it is necessary to switch between networks to ensure high-speed and continuous communication. When performing vehicle optical communication in an outdoor environment, it is easy to cause a ping-pong effect due to frequent network switching, which will cause unnecessary switching delays, as well as wrong switching judgments and network stability. Therefore, to achieve high-speed and stable visible light communication on the internet of vehicles, it is necessary to analyze the communication and network-switching characteristics of the network and design a reasonable switching algorithm.

Methods This study investigates the problem that the evaluation gap between networks is too small when the vehicle is switched between networks in the visible light communication system of the outdoor car network. It can easily cause a ping-pong effect. It is proposed based on the analysis of the network-switching characteristics. An improved multiattribute handover decision algorithm is proposed. First, we classify and process the main attributes that affect communication performance in different networks, input the processed attributes, and forward the attribute indicators. Then, we use the improved objective weight solving algorithm to correct the attributes after normalization. The indicators are processed to obtain completely independent and objective network selection weights in the outdoor internet of vehicle environment to reduce the influence of the dependence among different attributes on the network selection. Finally, the subjective weights of the network selection are obtained by combining them with objective weights. Based on the consistency of subjective value, objective value, and optimality of the evaluation results, an improved subjective and objective weight distribution model is proposed. The final weights of each attribute are obtained according to the distribution model, and the weights are used. After calculating the network evaluation value of each network, we sort and select the network with the largest evaluation value as the target network to obtain the correct handover decision.

Results and Discussions As a comparison, we take the number of vehicles and speed as the influencing factors, and compare it with the traditional analytic hierarchy process (AHP), the switching algorithm based on the decision tree (DT), and the switching algorithm based on principal component analysis (PCA). Using the handover failure rate, average bit error rate, and network throughput as indicators, the proposed algorithm is validated and compared in the simulation environment (Table 1). The simulation results show that under different conditions, the handover failure rate [Fig. 3(a)] and average bit error rate [Fig. 3(b) and Fig. 4(a)] are reduced and the network throughput [Fig. 4(b)] is increased when the number and speed of vehicles increase. Therefore, the proposed algorithm can effectively determine the best target network in a more complex environment, such as more communication users, higher frequency handover requests, and faster vehicle speeds, and enables the system to have higher communication performance.

Conclusions In this study, the algorithm design and simulation demonstration of the above processes are conducted. The simulation results show that the proposed algorithm can determine the best target network. Compared with the three algorithms of AHP, DT, and PCA, the proposed algorithm reduces the failure rate of handover between networks and effectively avoids the ping-pong effect. Therefore, the proposed algorithm can effectively implement and improve system performance when external parameters that affect communication performance change. The multiattribute switching decision-making scheme of the visible light communication heterogeneous network based on the internet of vehicles in the outdoor environment provides an effective reference method.