and the latest information is immediately processed and passed on (not held back).
As [12] correctly points out, the approach
is memory (buffering) and time (reprocessing) consuming.
We found that, in our case, this aspect could be neglected
because the fusion region and thus the number of potential
fusion objects is small. A short summary of algorithms for
the OSMM problem is given in [12]. [6] discusses different
synchronization strategies to minimize the worst case latency.