Wireless sensor network (WSN) is a core component of multiple smart city applications. Utilizing the same WSN for multiple applications helps reduce cost. However, satisfying quality of service requirements of these independent applications is very challenging. For instance, uncoordinated path selection for data dissemination may result in the formation of queues in the WSN violating end-to-end delay requirements of several applications. To this end, we propose a software defined network based approach to ensure satisfaction of individual delay constraints while ensuring minimal increase in the average queue length of the WSN. The approach utilizes a logically centralized controller to generate a comprehensive view of the whole network in a scalable manner. We develop several graph theoretic algorithms to reduce the number of nodes and edges in the communication paths and to identify the most suitable communication paths for each application so that end-to-end delays are minimized. The evaluations demonstrate that our approach performs up to 34% better than existing works and up to 14% worst in comparison to the optimal solution for different topologies, network sizes, and end-to-end delay requirements. Moreover, performance of the proposed graph theoretic algorithms is also measured w.r.t. time.