Open Source Solution Required: Issuing 100 million secure voter ID cards to the citizens of Bangladesh

[Raqueeb Hassan]

Hello,

Given the idea of having voter ID card for all the citizens of
Bangladesh, I was thinking of assessing few things before it actually
starts. The election commission, the government agency responsible for
issuing voter ID cards in Bangladesh might plan to use this - not only
to hold a fair election but to facilitate its citizen to establish
their credentials seeking access to all government and other
commercial privileges - from opening a bank account to buying a cell
phone, etc.

Project: Cost Assessment of Data center for Voter List supported by
Biometrical Technology.

It might sound pretty lame, but I was trying to figure out what might
be the approximate cost when it comes to assess the required
computational power for satisfactory finger print matching speed and
the storage involve of 100 million records. I know this complete
solution has to be multi-platform oriented, but it is always preferred
that a single solution developer can optimize the requirement of
number the clusters involve for fingerprint matching at a ratio of
1:100 millions. The effective matching speed increases proportionally
to the number of cluster's nodes, which we are yet to explore. The
COTS hardware can be used for matching AFIS. If google can do it, we
should be be able to reach that target.

Here are some details:

a. # of fields including image and tenprint information = 28 = 100 KB
b. Maximum byte in WSQ or other biometric standard (tenprint)= 4.2 KB
x 10 = 42 KB
c. Image = 20 KB

Total size = 162 KB/Person.

We have 100 million people to be issued with voter ID card.

<can be skipped>

Few of the Existing Solution Provider's case study (through my web
research)

Digimarc, a company based in US has similar solution on photo-based
voter ID card, which was applied to countries like Mexico and Haiti.
Hyundai has more advanced solution via its unique AFIS interface. It
does data conversion, processing and searching (1:1, 1:N) as per its
brochure. Neurotechnologi ja from Lithuania has great product like
Megamatcher that is designed to meet large-scale biometrical
identification and verification needs. The SDK provided for this makes
it very attractive for AFIS developers to have solution like this.
Say, the example of HANIS can be cited, which will eventually become
one of the largest civilian fingerprint databases in the world. For
better identification of 43 million citizens, the department of home
affairs in South Africa (HANIS) contacted Unisys along with other
partners. The government of Malaysia's "MyKad" is another example.

The case study of EURODAC is one of the best examples of AFIS. As it's
says "under the EURODAC system, each participating State will promptly
take the prints of all fingers of every asylum seeker over the age of
14. These fingerprints are compared with fingerprint data transmitted
by other participating States and already stored in the central
database. If EURODAC reveals that the fingerprints have already been
recorded, the asylum seeker will be sent back to the country where his/
her fingerprints were originally recorded."

For storage solution; I think Netapp's "data ontap GX" systems are
optimized to meet the demanding requirements of secure voter ID cards
database which will eventually lead to the formulation of smart
National ID system. It's pricey; do we have cheaper alternatives?
Oracle is the greatest database on earth but can that be cheaper (CAL)
for developing country like ours? Mysql, localization of databases and
use of clusters can serve our purpose.

<skipped end>

These solutions might be little out of hand when budget is tight.
Tough call; but we would like to do this mammoth job utilizing our own
local resources from the global perspective. With proper guidance, I
believe, we can do that too. The open source solution can be
prioritized when that outsmarts other competitive commercial
solutions. We will have a separate project for data accumulation which
is kept aside here.

Any help is appreciated on assessing of financial aspect of the said
data center.
Thanks in advance.

--
Raqueeb Hassan
Bangladesh

[Chris Davies]

Raqueeb Hassan <wi*******@gmai l.comwrote:

Given the idea of having voter ID card for all the citizens of
Bangladesh [...]

It might sound pretty lame, but I was trying to figure out what might
be the approximate cost when it comes to assess the required
computational power for satisfactory finger print matching speed and
the storage involve of 100 million records. [...]

Take a long hard look at the UK Government's similar project for ID
Cards. The scope and complexity are being reduced on an almost monthly
basis.

Chris

[Raqueeb Hassan]

On Feb 20, 10:16 pm, Chris Davies <chris-use...@roaima.c o.ukwrote:

Raqueeb Hassan <widean...@gmai l.comwrote:

Given the idea of having voter ID card for all the citizens of
Bangladesh [...]
It might sound pretty lame, but I was trying to figure out what might
be the approximate cost when it comes to assess the required
computational power for satisfactory finger print matching speed and
the storage involve of 100 million records. [...]

Take a long hard look at the UK Government's similar project for ID
Cards. The scope and complexity are being reduced on an almost monthly
basis.

Chris

Thanks for the pointer. If I can get your message right, it is
possible to do such kind of matching solution within a month. Can I
outsource the whole or part of the program? Can you please link me to
that UK solution, at least how it was done as a project?
--
Raqueeb Hassan
Bangladesh

[Chris Davies]

Raqueeb Hassan <wi*******@gmai l.comwrote:

Thanks for the pointer. If I can get your message right, it is
possible to do such kind of matching solution within a month. Can I
outsource the whole or part of the program? Can you please link me to
that UK solution, at least how it was done as a project?

It's ongoing and it's (apparently) a nightmare for all concerned. A
quick google for "UK Identity Card" brought up a significant number of
relevant links.

Chris