Inhabitants Per Household Drives Bandwidth Demand, Study Finds
Number of users per household is likely to emerge as the key driver of bandwidth demand, according to consultants Robert Kenny & Tom Broughton, in a study conducted for the U.K. Broadband Stakeholders Group.
A single-person household in 2023 will require require only about 5 Mbps and 8 Mbps to 10 Mbps for shorter periods of time. Also, the broadband connection is idle for most of the time.
A multi-person household with four occupants using high-definition streaming will experience appreciable usage almost constantly during the busy hours, with approximately 90 minutes of
demand of 25 Mbps or more per month, the study suggests.
Usage of video streaming, and especially use of higher-definition video sources, also will affect bandwidth consumption, but the key factor will be number of users per connection, one might reasonably conclude.
Looking across all households, the model indicates that the median household will require bandwidth of 19 Mbps by 2023, while the top one percent of high-usage households will have demand of 35 Mbps to 39 Mbps.
Those forecasts will strike some as low, given expectations that many consumers will have access to much-higher speeds by 2020 or perhaps 100 Mbps or higher.
But the analysts note that 64 percent of U.K. households are composed of just one or two people, limiting the effective amount of required bandwidth.
For example, even if two people are each watching their own HDTV stream, each surfing the web and each having a video call all simultaneously, the total bandwidth for this use
case is 15 Mbps in 2023.
Also, the growth of video consumption will be matched, to some extent, by improvements n video compression techniques that will reduce required bandwidth by about nine percent annually, for standard definition, high definition and 4K TV alike.
In addition, an increase in traffic does not necessarily equal an increase in maximum bit rate requirements. Up to a point, higher usage can occur without necessarily requiring an upgrade of top speeds.
The other issue is that the crucial parameter is peak usage, not average usage, which tends to be quite low, across a whole network. Even when data consumption per connection, over a month, is 23 GB, most of that consumption occurs in a “spiky” manner.
About 34 percent of monthly consumption happens in the 6 pm to 12pm period. During those “busy hours,” data consumed per connection is 7.8 GB.
Traffic per hour during the busy hours is 43.4 MB, but average usage is just 0.10 Mbps. Likewise, average modem sync speed is 12.7 Mbps, while average utilisation of the network is just 0.9 percent.
On the other hand, the study deliberately excluded the top four minutes a month of usage (peak demand) to get a better sense of sustained or typical demand. Accommodating the absolute four minutes of peak usage would boost a four-user household speed threshold up to 50 Mbps.
The authors also note that changing end user expectations could significantly affect supplier requirements. For example, For example, the analysis assumes users will tolerate 10 minutes waiting time for a console game to load. If that load time was reduced to 2.5 minutes, then 16 percent of households would require 83 Mbps.
Reducing the waiting time further would quickly take demand over 100 Mbps for those households.
Perhaps wisely, the study notes that “one cannot predict the future with exact certainty.” So the study conclusions do not include any impact of demand stimulation by providers or potential new applications that could boost demand for faster connections.
“In some cases we believe that current usage was constrained by current bandwidth, rather than reflecting what might be reasonably expected absent this constraint,” the authors note. In other words, actual future levels of demand could vary significantly from past consumer behavior.
That is key. Economists couch their conclusions using an important qualifier, “ceteris paribus” (all other things being equal). Of course, when new applications, new devices, new retail offers, new bandwidth and new access technologies become available, they change existing behavior. So behavior in the real world tends not to reflect “ceteris paribus.”
As with the Heisenberg Principle (often called the “uncertainty principle”), which stipulates that, when attempting to measure a particle’s position, the more precisely the position is determined, the less precisely the momentum is known in this instant, and vice versa. In other words, one can know where a particle is, or its momentum, but not both with equal precision.
A related “observer effect” is probably more germane. The problem, in essence, is that the act of measurement changes the process or thing being observed. The reason is that use of measurement instruments necessarily changes the quantity of the measured process or object.
An easy example is the use of a tire pressure gauge to measure tire pressure: applying the gauge lets some air out of the tire, changing the status of tire air pressure somewhat.
For Internet apps, almost anything “new,” ranging from an app or device to the way retail services are priced and packaged, can change user behavior.