Performance of Dynamic-Frame-Aloha protocols: Closing the gap with tree protocols

The Dynamic-Frame-Aloha protocol, largely studied in the 60s in the field of random access satellite systems, is nowadays commonly applied also to Radio Frequency Identification systems to orchestrate the transmissions from the tags to the reader. In a nutshell, tags respond to reader's interrogation in slots randomly chosen in a frame whose size is dynamically set by the reader according to the current backlog (remaining tags to be resolved). In this paper, we explore the performance of the DFA protocol under Poisson-distributed population of tags when different strategies are adopted in setting the frame length and estimating the traffic backlog. We further analytically characterize the best strategy in the two cases where the frame is entirely explored or a new frame can be restarted at any slot.