Data communication in wireless sensor network

Several measurements has been done:
 * End-to-end throughput for 1-hop, 2-hop and 3-hop flows.
 * End-to-end delay for 1-hop, 2-hop and 3-hop flows.

Future measurements:
 * Shadowing effects.
 * Multipath interference.

If we have our own routing protocol, we can measure its performance in these scenarios. But we are just using fixed routes or AODV. I do not see much point for doing all these above measurements. What conclusion can we achieve?

There is one feature of the SensorNet node, acting as cluster heads. Sensors report readings one-hop to SensorNet node, which will archive the readings locally. Then there are two ways to deliver the data: (1) SensorNet node periodically sends readings back to the control center; (2) Control center sends queries to some nodes, and then nodes send relevant data back.

\Under this two-layer architecture, the query will not consume too much resource comparing to the network where data could be stored in every sensors. So which way is better? Maybe for some application, there is only one option.