Some thoughts on WSN routing protocols, and prob facing

I'm currently working on my graduation thesis about routing protocols
of Wireless sensor network. Proposed by researchers from USC/ISI, a
protocol called "Directed Diffusion" is one of the most widely known
sensor net protocols. Although it was originally implemented on real
hardware testbed(Lucent WaveLAN), it has been migrated to the network
simulator 2 some years ago.

I have been reading the C++ code and Tcl scripts the last two weeks.
I've also been testing results for some simple scripts in the test
suite. It works, but the implementation in NS-2, diffusion 3, is far
from mature. The diffusion mailing list of USC/ISI has been quite for
years, and seems no one is doing research based on DD anymore, partly
bcoz the hard-to-use and immature implementation of NS-2 version of
DD. As the consequences, many researchers turn to some other kinds of
protocols such as LEACH, a cluster-based network protocols.

What I have in mind is to design a multi-tier sensor network protocols
with the help of multi-sink cooperation. There is too much
concentration on multi-source cooperation, or data aggregation, data
fusion etc. But few attention on multi-source condition. Since last
year, there have been few proposals about multi-sink data gathering,
but still some mathematical stuff. The work is done mainly by
mathematicians, not network specialists. And multi-sink cooperation is
of very much importance, especially when remembering the
application-specific nature of sensor network protocols.

Let's see what can I coin out. But before that, I need to carry out a
better way rather than solely depends on NS version DD. In the current
situation, it's impossible for me to reach a real testbed like the one
the guys in USC/ISI are using.