The paradigm of multi-agent systems has been actively applied to cloud computing to provide cost-effective, decentralized problem-solving. Rather than processing the tasks that require enormous computing resources locally, they can be processed in the cloud. In this research, we present how these two paradigms can be effectively used to handle emergency services in any metropolitan transportation system. In the proposed architecture, agents are placed at signal nodes to collect data and send it to cloud services. These agents have the ability to control the signal nodes in an emergency. They also communicate emergency information between several nodes. The cloud service is the system’s backbone and employs heterogeneous agents to manage emergency services. It analyses the input data and determines the presence of emergency services. The main contribution of the research is to utilize a priority-based mechanism for handling exigency vehicles rather than using traditional timer-based signals. In addition, we provided case studies demonstrating the usefulness of the proposed framework.