This paper considers distributed protocols for assigning IDs. Normally the maker of an RFID system would simply assign each ID sequentially in the factory, or the user would manually burn it in to EPROM. But it may be cheaper and simpler to manufacture units identically, and build the capability to become unique into each one. I describe three distributed protocols which make use of hardware random number generation to create unique IDs. In the first, trivial scheme, random IDs of sufficient length are independently chosen by all units. This requires no communication among the units, so it could be used to assign identities in circumstances in which central coordination or communication among the units is impossible. However, because no communication or central coordination is used, the IDs may be quite long. I then present a second protocol which requires a modest amount of communication to assign a set of unique, short IDs. Finally, I present a protocol with more substantial communication requirements that accomplishes the same ID assignment task in much less time.
[1]
Lenore Cowen,et al.
Efficient asynchronous distributed symmetry breaking
,
1994,
STOC '94.
[2]
Edsger W. Dijkstra,et al.
Solution of a problem in concurrent programming control
,
1965,
CACM.
[3]
Russ Bubley,et al.
Randomized algorithms
,
1995,
CSUR.
[4]
Arvin Park,et al.
The Processor Identity Problem
,
1990,
Inf. Process. Lett..
[5]
C. Hopkins.
Principles of electrolocation and jamming avoidance in electric fish: Studies in brain function. Vol. I By W. Heiligenberg. New York: Springer Verlag (1977). Paper: $10.40
,
1978,
Animal Behaviour.