This is 'standard queuing theory'. If you like that kind of thing
check out Kleinrock's texts on the subject.
If you don't then:
1) All queuing theory results are HEAVILY dependent on the assumptions
about traffic patterns.
2)The usual assumption is that the traffic is "Poisson" in both
arrival rate and the time it takes to serve one arrival. (Most benign
assumptions)
Under these assumptions the theoretical throughput of an ALOHA scheme
is 1/2e of the 10MPBS. A slotted ALOHA doubles that to 1/e of the
10MBPS (e=2.71818)
Note that slotted Alohoa sort of needs all the 'remotes' to be
synchronized, not always possible in practice.
3) Since 1/e (38%) is not great throughput, this has led to schemes
like Carrier Sense (CSMA and CSMA-CD) that most 'wired' ethernets use
these days.
4) If your traffic is not Poisson, and in particular if it is
multimodal, i.e. you have lots of short requests such as web pages and
also frequent long requests such as ftp transfers, then the above
numbers may not hold, (generally things get worse) and the math
becomes much more complex. |