Thursday, May 21, 2015

Connected Car Signaling, Not Bearer Traffic, Will be an Issue

Mobile network planners necessarily must grapple with problems fixed network architects traditionally have not faced, namely the mobility of users across the network. In other words, even if radio locations are fixed, usage is more dynamic than on a fixed network.

For that reason, signaling overhead is a bigger part of mobile network planning than of fixed network design and management.

A recent Qualcomm study of  smartphone application traffic in background mode illustrates one element of the problem. That study found some news applications generate four to six requests per hour, compared to social networking applications which generate one to four per hour, and location based applications which generate two to three per hour.

Bandwidth isn’t so much the issue as the signaling operations. A specific weather application might connect for less than three seconds, uploading less than two kilobytes of data. So it isn’t the bearer traffic; the impact comes from use of radio resources to create and tear down a session.

Social networking applications connected for two to four seconds and uploaded one to three kilobytes of data.

If nothing else is done (and obviously, something will be done), connected cars at rush hour could double data traffic double in certain cells, researchers at Machina Research say. And the issue there might be similar: intensive signaling, more than actual bandwidth consumption.

The study, commissioned by analytics company TEOCO, predicts a 97 percent increase in data traffic over the next decade, caused primarily by connected cars.

Some of you would not be surprised by a prediction of 100 percent increase in traffic, on any network, at any site, over a decade.

What the study intends to highlight is the specific new demand created solely by the connected cars.

By 2024, Machina Research predicts “machine-to-machine” mobile network connections increase from 250 million in 2014 to more than 2.3 billion worldwide.

Obviously, all those new devices, an order of magnitude more, would produce, all other things being equal, an order of magnitude additional demand. But the usage profile might be notable for its difference from human-used smartphones, tablets and PCs.

M2M applications and services will account for just four percent of overall network traffic in 2024, Machina Research predicts.

So bandwidth consumption is not the problem so much as network resource management, which always is more complicated than comparable fixed network planning, since the nodes are stationary.

The study suggests a number of ways the management problems can be addressed, including the integration of capacity supplied by unlicensed networks and offload.

History suggests that awareness of the coming problem will lead to actions that prevent the problem from developing.

No comments:

AI "Performance Plateau" is to be Expected

There is much talk now about generative artificial intelligence model improvement rates slowing. But such slowdowns are common for most--if...