Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Options for Processing Mobile Phone Data
Margus Tiru
Positium
What is Positium?Methodology and technological platform for processing Mobile Big Data for human mobility monitoring, analyses and statistical indicators
Positium Data Mediator
The indicators can be used in population statistics, tourism management and marketing, spatial planning, transportation modelling, spatial marketing, safety and security analyses, and other domains
Our solutions use anonymised Mobile Big Data from Mobile Network Operators (MNO) to turn billions of location points into meaningful and understandable statistical indicators, analysis of
decisions, smart city solutions, and research
How to get from Raw Data to Results?MNOs collect data and often process for their
internal purposes
The (residual) data is highly valuable in variety of domains (population, mobility, tourism, etc.)
Legal, administrative, technical and business aspects are difficult
Private data owned by MNOs (data controller, responsibility before the customers)
Asset for MNOs (there is value in data)
Applications of Mobile Data Analytics Population statistics
Everyday commuting
Transportation planning
Urban planning
De facto population
Tourism statistics
Epidemiology
Ecological aspects of human mobility
Economics
Proximity of population to risks
Safety and Security
ICT indicators (measuring the Information Society)
Spatial marketing
Scientific research in many areas
Tiers of the Processing
Tier I Extraction
Tier II Processing
Tier III Aggregated
results
Two Main Options
Distributed Processing Centralised processing
Distributed (In-House) Processing
Data handled by MNOs
From extraction to aggregation and indicators
Big Data departments, commercial customers
NSI can merge the results from several data providers
Distributed (In-House) Processing
Pros
Privacy protection aspects “resolved”
Less skills needed for NSI
MNOs use the data also for commercial services
Less expensive for NSI (unless NSI has to compensate)
Cons NSI does not have hands on
raw data – live with what you get (e.g. cross-roaming for tourism)
The methodology might be different (unless it is OPAL type – bring algorithms to data approach)
No control over processing
Burden on MNOs = requires compensation (paying for the data)
More expensive in total
Centralised (NSI) Processing
MNO: Extraction of the data and transmission to NSI
Full processing chain by NSI
One methodology, trusted and controlled results
Traditional way
Centralised (NSI) Processing
Pros
Full control over methodology and algorithms
Trusted, controlled results
No burden on MNOs
Least expensive in total
Cons
Privacy protection
Requires technical resources and skilled employees
More expensive for NSIs directly
Centralised (Trusted Third Party) ProcessingMNO: Extraction of the data and transmission to
the Trusted Third Party
Third Party can be govt. organisation (TRA, Ministry of ICT) or
private organisation (University, Company)
Centralised (Trusted Third Party) Processing
Pros
Defined one methodology (?)
Almost controlled results
No burden on MNOs
Less skills needed for NSI
Less expensive in total
Cons
Unknown methodology (unless it is OPAL type –bring algorithms to data approach)
Privacy protection
More stakeholders = more complicated business model
Which Is the Best?
Technically, all options are possible
Depends on the business model and regulations
Market Conflict There are two different markets for the data which overlap
Public Interest
Private Market
Conflict examples: Customers for NSI / MNO
NSI MNO
National-level Balance of Payments travel item Yes No
Municipality-level tourism visitation statistics for tourism boards Yes Yes
Shopping centre visitor footfall, catchment area, competitor location analysis
No Yes
Nation-wide everyday commuting data for monitoring the fuel consumption indexes to actual mobility of the population
Yes No
Municipality public transportation OD-matrices for optimisation No Yes
National carrier (e.g. Air France) route optimisation data Yes Yes
Outdoor media location analysis based on daily population (de facto population grid) and regular routes, repeating sighting
No Yes
Market Conflict Agree on the division between markets
Public Interest
Private Market
Why isn’t there a working solution everywhere already?No real solution has been proposed and agreed
upon
NSIs are too conservative and slow?
Budget cuts in public domain result in decrease of investment in innovation
Regulations of EU do not provide possibility to leave grey area – no clarity
There are no really usable mathematical methods for 100% anonymization that preserve the value of the longitudinal data
Barriers
We have already established that there is a need in the NSIs of many countries to start using mobile data
Here are the top barriers to big data adoption: Identify funding
Engage policy-makers
Identify key partners and get access to data
Skills, capacity building
TechnologySource: UNSD GWG on Big Data and Official Statistics, Big Data survey among NSIs
What Is Required To Make It Happen?
Willingness to Invest
That also means finding the source for financing the innovation
Also seeing the actual ROI for the society in long term
Government, data owners, users, people
Leaders
Eurostat. UNSD. OECD. etc. to set the legal and methodological framework and lead the initiatives
NSIs who would take decisive steps and invest
MNOs willing to participate, play globally and see the opportunities
Skills“By combining a scientific background with computational and analytical skills, you can put yourself a ‘cut above the rest’”
- Big Data is the new black
But you can distribute these skills, train the existing people and cooperate with Universities and other partners (MNOs)
Common Methodology
Methodology exists,
it is not finished
but there is plenty to start with
Technical Tools: Algorithms
There is obvious need to use the same methodology and same algorithms
There are several companies, universities, and organisations who have developed the methodology and algorithms already (e.g. Positium )
OPAL
DIY?
Positium and University of Tartu
have spent a combined total of
more than 450 months developing
the methodology and algorithms
Technical Tools: IT infrastructure
Computational and storage resources
Distributed processing within cloud computing environment
Mostly MNOsalready have thesetools
Business Model
Partnership model with data owners (PPP?)
Incentives for data owners? Cash (pay for data)?
Reimbursement for implementation (pay for technical service of providing the data)?
Provide other benefits
Cooperation with other govt. users How about partnering with Communication Regulators?
ITU Big Data Indicators for ICT
BD01 – Percentage of the Land Area Covered by Mobile-Cellular Network
BD02 – Percentage of the Population Covered by a Mobile-Cellular Network
BD03 – Usage of Mobile-Cellular Networks for non-IP Related Activities, by Technology
BD04 – Usage of Mobile-Cellular Networks for Internet Access, by Technology
BD05 – Number of Subscriptions with Access to Technology
BD06 – Active Mobile Voice and Broadband Subscriptions, by Contract Type
BD07 – Average Number of Active Mobile Subscriptions per Day, by Contract Type
BD08 – Active Mobile Devices
BD09 – IMEI Conversion Rate
BD10 – Fixed (Wired) Domestic Broadband Capacity, by Speed, Contract Type
BD11 – Mobile Domestic Broadband Capacity, by Speed, Contract Type, Technology
BD12 – Mobile International Broadband Capacity, by Contract Type
BD13 – Inbound Roaming Subscriptions per Foreign Tourist
BD14 – Fixed (Wired)-Broadband Subscriptions, by Technology
BD15 – Fixed (Wired)-Broadband Subscriptions, by Speed
NSI TRA
Gov.
€
Regulations and Privacy Protection In EU this is the biggest obstacle
It requires more work and change in EU and national regulations
However, it has been stated in several countries by DPAs that if the processing was done in-house by MNOs, and the identifiable data was deleted after the results – it would be acceptable
Extraction ProcessingAggregated
results
So What Are We Waiting For?
Related Documents
