Tikalon Blog is now in archive mode.
An easily printed and saved version of this article, and a link
to a directory of all articles, can be found below: |
This article |
Directory of all articles |
Cellphone Census
November 24, 2014
I worked for a while in
broadcasting during the
1960s. It was an interesting time, since I was introduced to
personality types that aren't found in an
academic or
scientific environment. Those were the days of "
top forty,
news,
weather, and
sports," and the on-air personalities were interesting characters for the simple reason that they needed to be or they wouldn't have been hired.
At that time, the
advertising manager had an interesting idea on how to do a
survey to determine the relative popularity of the
AM radio station in the
local market. His idea, which he likely read in one of his
trade publications, was to see what stations were tuned-in on the
radios of
parked cars around the
city. This was easy, since there were just four radio stations at the time. Today, that type of survey couldn't be done for two principal reasons. First, people would wonder about people peering into every
parked automobile on a
street; and, second, modern radios show no indication of what station was tuned when the automobile is off.
In those days, however, radios were
mechanically-tuned with a
knob that also moved a
frequency indicator along a slide display (see photo). Since the
FCC assigns station frequencies at intervals to prevent
interference between stations, it was easy to see from this indicator which of the four stations (950, 1250, 1310, and 1550
kHz) was last heard. Later, car radios were designed to receive both AM and
FM radio stations, the tuning dial was used for both, and it would have been hard to determine whether an AM or an FM station was last tuned; but, this survey method was workable at that time. Our station ranked first in this survey.
An old analog car radio receiver with a slide tuning indicator. This is tuned to 1440 kHz on the AM band, but you can tell that it isn't a radio for use in the US, since the FM band extends from 76-90 MHz, the Japanese FM band, and not 88-108 MHz US band. (Via Wikimedia Commons.)
Wireless technologies, such as
RFID, have enabled very efficient
inventory control. This is, in effect, a survey of
goods, but other wireless devices enable surveys when that's not their intended purpose. One controversial method of
traffic survey is through use of the now ubiquitous
E-Zpass system that's intended for
electronic toll collection. Since E-Zpass RFID tags can be detected at non-toll locations, they're also used to provide estimates of travel time between points. While it's claimed that these data are scanned in an
encrypted form and deleted as soon as the travel time estimate has been completed, people are still concerned about later "
enhancements" to these systems.
Nearly everyone has a
cellphone, and this includes residents of
Africa, who use cellphones because there are few
landline telephone. The
technology exists to
track cellphones, at least at the
granularity of the
cell tower grid, since cellphones transmit an unique
identification number along with their
voice and
data signals. Such tracking has been used to assemble data on
population movements in Africa that might predict how the
Ebola virus would spread.[1]
Orange Telecom, a
West African mobile network operator, has provided cellphone location data for
Senegal, and it previously provided such data for the
Ivory Coast. The Senegal data was from 150,000 phones in 2013, it was
anonymized and
aggregated, and then
analyzed for population movement by
Flowminder, a
Swedish nonprofit organization.[1] Although not yet used in the Ebola campaign, the data do show where people go after leaving an Ebola hot spot, thus suggesting where the
disease will appear next. As in the E-Zpass example above, such tracking could reveal
social and business connections of individuals, so privacy problems do exist.[1]
Can you locate the Sahara Desert?
The population density of Africa, as published by The WorldPop project (worldpop.org.uk).
(Image from The WorldPop project, released under the Creative Commons Attribution 4.0 International License.)
Scientists from the
Université catholique de Louvain (Louvain-la-Neuve, Belgium),
Northeastern University (Boston, Massachusetts), the
Fonds National de la Recherche Scientifique (Brussels, Belgium), the
Université Libre de Bruxelles (Brussels, Belgium), the
Université de Lorraine (Vandoeuvre-lès-Nancy, France), the
University of Louisville (Louisville, Kentucky), the
University of Southampton (Southampton, United Kingdom), the
National Institutes of Health (Bethesda, Maryland), and the
Flowminder Foundation (Stockholm, Sweden), have recently used cellphone data as a means for
census-taking. Their study showed that such estimates compare favorably to data from traditional techniques.[2]
The usual census-taking method, at least in the US, involves
mailed questionnaires, and home visits to those who don't respond. All this could be done more quickly, although with less
accuracy at first, using cellphone data. However, in the extended period between traditional census-taking, it would offer better accuracy. The international
research team, led by
geographer,
Catherine Linard, of the Université Libre de Bruxelles and
data scientist,
Pierre Deville, of the Université Catholique de Louvain used cellphone data as a way to estimate the population density of
France and
Portugal.[3] They used a
dataset of more than a billion call records from these countries.[2]
In the case of Portugal, the call records were for two million users, which is about 20% of the population. For France, the records were for seventeen million users, which is about 30% of the population. Call records for Portugal included the cellphone identifier code, the locations of the originating and receiving cellphone towers, and the start and stop times of the call. The data for France were limited to just the day of the call and the tower locations.[3] Corrections were made to allow for the fact that cell towers are not
uniformly distributed.
Not unexpectedly, the population distribution varied during time of day, time of week, and
season of the year; e.g.,
holiday and
vacation time away from
work is generally
sacrosanct in France. More people were in cities during the work week, and there were more people in
rural areas on weekends.[3] The
authors remark that
texting might be more popular in some countries than voice communication, and that would modify the data.[3]
Politicians are not about to relinquish their fortunes to
electoral districts apportioned by cellphone usage. However, these data would be useful in countries where traditional methods of census-taking are unreliable. One example cited in a
Science article is that the last census of the
Democratic Republic of the Congo took place in 1984.[3]
French population density, Sunday (left) vs Monday (right). Paris is the large upper blotch. (Still images from a YouTube Video.)[4)]
Possibly the most important application of such cellphone data is the ability to track population flow in
emergencies to allow for adequate impact assessments and intervention planning.[2] As the authors write in their article,
"...The prospect of being able to map contemporary and changing human population distributions over relatively short intervals exists, paving the way for new applications and a near real-time understanding of patterns and processes in human geography."[2]
References:
- David Talbot, "Cell-Phone Data Might Help Predict Ebola's Spread," Technology Review, August 22, 2014.
- Pierre Deville, Catherine Linard, Samuel Martin, Marius Gilbert, Forrest R. Stevens, Andrea E. Gaughan, Vincent D. Blondel, and Andrew J. Tatem, "Dynamic population mapping using mobile phone data," Proc. Natl. Acad. Sci., published ahead of print, October 27, 2014, doi:10.1073/pnas.1408439111. This is an open access article with a PDF file available here.
- Jia You, "Taking the census, with cellphones," Science, October 27, 2014.
- Pierre Deville, "Dynamic Population Mapping Using Mobile Phone Data," YouTube Video, October 27, 2014.
Permanent Link to this article
Linked Keywords: Broadcasting; 1960s; personality; academia; academic; science; scientific; environment; top forty; news presenter; weather forecasting; sports commentator; advertising manager; opinion poll; survey; AM broadcasting; AM radio station; local market; trade magazine; trade publication; vehicle; radio; car radio; parked; automobile; car; city; street; mechanical; control knob; frequency; Federal Communications Commission; FCC; electromagnetic interference; hertz; kilohertz; kHz; FM broadcasting; FM; analog electronics; AM band; United States; US; FM broadcasting; FM band; megahertz; MHz; Japanese; Wikimedia Commons; wireless technology; radio-frequency identification; RFID; inventory control; goods; traffic; E-Zpass; electronic toll collection; encryption; encrypted; information privacy; mobile phone; cellphone; Africa; landline telephone; technology; mobile phone tracking; granularity; cell site; cell tower; grid; identifier; identification number; voice; data transmission; data signal; population; Ebola virus; Orange Telecom; West Africa; mobile network operator; Senegal; Ivory Coast; data anonymization; anonymize; data aggregator; aggregate; analysis; analyze; Flowminder; Sweden; Swedish; nonprofit organization; disease; social network; social and business connection; Sahara Desert; population density; Africa; worldpop.org.uk; The WorldPop project; Creative Commons Attribution 4.0 International License; scientist; Université catholique de Louvain (Louvain-la-Neuve, Belgium); Northeastern University (Boston, Massachusetts); Fonds National de la Recherche Scientifique (Brussels, Belgium); Université Libre de Bruxelles (Brussels, Belgium); Université de Lorraine (Vandoeuvre-lès-Nancy, France); University of Louisville (Louisville, Kentucky); University of Southampton (Southampton, United Kingdom); National Institutes of Health (Bethesda, Maryland); Flowminder Foundation (Stockholm, Sweden); census; census-taking; mail; questionnaire; accuracy; research; geographer; Catherine Linard; data science; data scientist; Pierre Deville; France; Portugal; dataset; uniform distribution; uniformly distributed; season of the year; holiday; vacation; wage labor; work; sacrosanct; rural area; author; text messaging; texting; politician; electoral districts; apportionment; apportioned; Science; Democratic Republic of the Congo; French; Sunday; Monday; Paris; YouTube Video; emergency.