Abstract: ‍ ‍In recent years， the communication-system demands of a vehicle-to-everything （V2X） network have included large-scale data transmission and stable communication connections. High-frequency communication has been introduced into the system， which has the characteristics of an ultra-high transmission rate and low latency. However， high-frequency communication is still troubled by the challenge of blocking. Therefore， as a technology to solve this problem， an intelligent reflecting surface （IRS） has become a hot topic for the high-frequency communication of a V2X network. An IRS is composed of a metamaterial， which contains multiple passive reflecting elements on its surface that are uniformly arranged to form an array. Moreover， the reflection coefficients of these elements can be adjusted in real time through an intelligent controller to control the direction and shape of the reflected beam. This makes it possible to intelligently manipulate the wireless transmission environment and provide services such as enhanced line-of-sight communication and extended communication coverage. The IRS was introduced to resolve the problem of the line-of-sight paths between the base station （BS） installed at the roadside and the vehicle users （VU） being randomly interrupted in the MU-MISO millimeter wave communication scenario of a vehicle-to-infrastructure network as a result of factors such as shielding and the high travel speed of the vehicles. Meanwhile， combined with the characteristics of this system， an alternate iterative optimization algorithm based on semidefinite relaxation （SDR） problems was proposed to improve the stability of the communication quality. This method decomposed the optimization of the base station beamforming matrix and IRS phase shift matrix into two subproblems， which were approximated as SDR problems by defining the semidefinite matrix variables and appropriately relaxing the specific constraints. Then， an iterative solution was obtained using the alternate optimization technique to realize the joint optimization， which ensured communication stability by maximizing the minimum signal-to-interference-plus-noise ratio （SINR） among the vehicle users simultaneously connected to the network. The simulation results showed that it significantly improved the SINR of vehicle users and sum-rate of the BS， ensuring stable communication between the vehicle users and BS in highly dynamic scenarios.