Directed diffusion is a prominent example of data-centric routing based on 
application layer data and purely local interactions. In its functioning it 
relies heavily on network-wide flooding which is an expensive operation, 
specifically with respect to the scarce energy resources of nodes in wireless 
sensor networks (WSNs). One well-researched way to curb the flooding overhead 
is by clustering. Passive clustering is a recent proposal for on-demand 
creation and maintenance of the clustered structure, making it very attractive 
for WSNs and directed diffusion in particular. The contribution of this paper 
is the investigation of this combination: Is it feasible to execute directed 
diffusion on top of a sensor network where the topology is implicitly 
constructed by passive clustering? A simulation-based comparison between plain 
directed diffusion and one based on passive clustering shows that, depending 
on the scenario, passive clustering can significantly reduce the required 
energy while maintaining and even improving the delay and the delivery rate. 
This study also provides insights into the behavior of directed diffusion with 
respect to its long-term periodic behavior, contributing to a better 
understanding of this novel class of communication protocols.