[LINK] SMS - shortcomings in emergency services

[Richard Chirgwin]

Linkers,

3G Americas has produced this white paper (PDF, you can skip the
registration), which provides the first serious, engineering-based
analysis I've seen so far:
<http://www.3gamericas.org//registration/register.cfm?content_id=118&goto=http://3gamericas.org/PDFs/Characterizing_the_Limitations_of_3rd_Party_EAS-Traynor_Sept08.pdf&newwin=0>

The conclusion, that SMS is unsuited to emergency notification, doesn't
surprise me. However, the detail of the analysis makes it well worth
reading.

In particular, the author notes that cellular networks are scaled for
high utilisation (so that they're profitable). That means a "peak load"
(same message to thousands of phones) is certain to swamp the network.
This, in turn, results in non-delivery of messages.

Also, the paper observes that it's nearly impossible for users to
authenticate an emergency notification - which means fake notifications
could be a significant problem.

Finally, the message surge would block voice calls - preventing (for
example) access to emergency service calls for any user within the
affected cell/cells:

>      Note that after the initial surge of emergency messages, nearly
> all messages
> and phone call requests cannot be delivered. In particular over 70% of all
> messages and phone calls are blocked. This value does not consider the
> impact of
> retransmission attempts by the SMSC for the text messages that were
> dropped
> during the initial surge. This result is also conservative in that it
> does not assume a
> continued increase of calls and text messages.

Richard Chirgwin