We’re trying to characterize Bluetooth connection latency for various scenarios. Ideally, we would like to produce a model that shows minimum, maximum and mean connection times. Initially, I'm not looking at BLE but if you have pointers there they would be appreciated.
For the actual connection, we’re are planning to start with raw RFCOMM for a connection made to a known MAC without paring (no security). Next, we would like to add SDP to this model. Finally, we would like to add device discovery.
Does anyone know what work is already available in this area (academic papers, parts of the standard etc.). Which parts of the standard are relevant here? How do the various tunable timing parameters affect this performance? What trade-offs would one see if they deviate from the standard values?
Now for the $6M question to finish. In your experience, how do the various BT implementations measure up to this theoretical ideal?
Any pointers you can offer will be greatly appreciated for our research. I'm expecting the result of our research to be fairly rigorous so we are not afraid of any of the math that will be needed.
If it matters, we are testing using both a WL1271L based TiWi-BLE along with BlueZ 5.X and also CC256x based module using Bluetopia.
Thanks,
Craig