Directed diffusion is a data-centric communication mechanism for sensor networks. It is supported by a flexible diffusion filter framework. GPSR (greedy perimeter stateless routing) is a routing mechanism commonly used in sensor networks. This project is an effort to connect GPSR with directed diffusion. Compared with existing routing mechanisms in the diffusion platform, including push, pull and gear, GPSR has quite different qualities in performance. It is an importation addition to the diffusion platform. As a widely known protocol, GPSR is also a good benchmark for ongoing and future research works. In this project, GPSR is implemented as a filter in the diffusion platform, and we performed simulations to compare the performance of different routing mechanisms. There are two findings from simulations results: a) overall GPSR is a rather efficient routing protocol. B) the behavior of GPSR is quite sensitive to node density. The performance will degrade significantly for sparse (barely connected) wireless networks. The poster will show simulation results with analysis. It will also show the internal structure and interfaces of the GPSR filter, as the implementation of the GPSR filter is good example of integrating new routing filters to the diffusion platform.