The prioritized random access mechanism has been researched in this paper. The current situation of large-scale node multiple access in the Internet of Things leaded to channel congestion and increased blocking rate. In M2M communication, there were a large number of nodes trying to RA. In this case, serious physical random access channel overload would occur, making it even difficult to access the network. Nodes spent a lot of time accessing the network, and even given resource management schemes could not guarantee acceptable end-to-end latency meet QoS guarantees. So, the traditional method has greater limitations in supporting different QoS in M2M communication, It would also have an unimaginable impact on the web, Therefore, it was necessary to propose a scheme to reduce the access delay while ensuring the success rate, The current priority-differentiated random access mechanism needed to consider how to match with the new features of M2M services. The reasonable random access scheme and efficiently manage the resource and allocated the channel were designed in the system where the traditional H2H service and the new M2M service coexist. The weights of different services were obtained by modeling and simulation. Therefore, by setting different threshold detecting preamble index matching service type and calling different back-off indexes to reduce the access collision, so as to optimize the random access mechanism. This study could make the node indicate its priority in the access process, and reduced the conflict and access delay. Under 5G communication technology, this solution would be widely used.