In the internet of vehicles (IoV), periodic beacon messages carry the function of traffic safety data interaction. However, when traffic density increases, these beacons will cause channel congestion. To solve the problem, an adaptive link parameters (ALP) assignment protocol based on predictable channel condition is proposed. The channel condition is estimated by traffic density, signal to interference plus noise ratio and channel busy ratio. According to the prediction of channel condition, beacon broadcast interval, power and modulation coding scheme are assigned. Congestion is avoided by controlling these link parameters. Performance of ALP is evaluated in the OMNET simulation platform. Results show that: 1. Packet delivery rate and latency are guaranteed in both congestion and idle channels. 2. Broadcast coverage and successfully received beacons number are improved in idle channel. 3. Channel utilization is high while avoiding channel congestion.