Increasing Offered Traffic Loading

What we really need most are applications that increase offered traffic loading in the
off-peak hours in a network. Figure 14.7 shows a 24-hour loading of a cell site in Biggleswade,
a rather sleepy suburban town in the United Kingdom (with the lowest
divorce rate in Britain). The vertical axis is Erlangs (voice traffic loading) and the horizontal
axis is time. Between midnight and six in the morning, Biggleswade sleeps
(probably why the divorce rate is so low), and so does the network. It is virtually
unloaded. At 8:00 A.M. Biggleswade wakes up and goes to work. At 1:00 P.M. Biggleswade
stops for lunch. Biggleswade starts going to sleep again as the afternoon progresses.
Asmall evening peak happens just about 8:00 P.M. as Biggleswade arranges the
traditional daily visit to the pub.

This is a pre-SMS plot of diurnal loading. The impact of SMS text messaging has introduced
additional loading in the evening. One text message may prompt multiple replies
(from the SMS buddy list), generating additional network traffic.. This loading includes
voice traffic stimulated by the SMS exchange. SMS is very economic in terms of bandwidth
used—a few hundred bits and relatively high value in terms of billability. This
means it has high perceived value to the user. Even a full-length SMS message only takes
1 1⁄2 seconds to send but can deliver a tariff gain equivalent to a 2-minute phone call.
SMS is store and forward, so not only do we have a low bandwidth requirement, we
also have low-cost bandwidth in terms of latency tolerance. This means memory bandwidth
allows us to use delivery bandwidth more effectively. In SMS, the traffic delay
may be just a few seconds. However, certain types of traffic may be delayed by several
hours and not suffer loss of value.
Video surveillance, for example, can be (although not always will be) delay-tolerant.
Provided sufficient buffer bandwidth can be made available in the device, then image
bandwidth can be delivered in the middle of the night for centralized (network-based)
storage. Similarly, if a handset has sufficient memory, it can be updated overnight, that
is, trickle fed by the network. The trick is to get the subscriber into the habit of charging
his or her phone overnight and leaving it switched on. The network can then
download, for example, the subscriber’s three most frequently visited Web sites. In the
morning, the user has the illusion of instantaneous access to delivery bandwidth; the
three sites are immediately available to the user. All the network has to do is difference
the site—that is, update any changes that have occurred. 356