This paper presents a decision support system for prediction of frozen soil thaw settlement coefficient to solve the possible thaw settlement problems affecting slope stability, which leads to rock pile section roadbed deformation, and threaten road safety. For this purpose, the prediction of frozen soil thawing settlement coefficient is decided using 5G communication. The prediction is constructed using MIMO-D2D system, increases the signal transmission dimension, improves the system throughput, and effectively allocates resources, which solves the problem of interference coordination between device-to-device (D2D) and cellular users in the multiple-input multiple-output (MIMO) system. A set of equipment pieces for in situ thawing tests of coarse-grained and giant-grain frozen soil is designed. Field tests are carried out to obtain the thawing coefficient of the frozen soil and calculate the thawing compression index, which can evaluate the influence of frozen soil thawing settlement on railway engineering. The experimental results of the research show that the elastic properties of frozen soil are weak, the unloading rebound is only 10% of the compression amount, and the rebound is concentrated in a short time. The experimental results show that the thaw settlement coefficient of frozen soil is 0.67%, which is classified as Class I according to thaw settlement, non-thaw settlement, and weak thaw settlement. The maximum thaw settlement at the slope of the tunnel entrance section is about 6.7 cm, which allows settlement after construction.