How Does Aadhaar Build a Secure Digital Identity?

TL;DR

Aadhaar gives each enrolled resident a random identification number after checking fingerprints and irises against the entire central database for duplicates. Its architecture combines online yes-or-no authentication, encrypted enrollment packets, horizontally scalable open-source components, multiple biometric providers, and a nationwide partner ecosystem designed to reach residents across India’s diverse languages, locations, and infrastructure conditions.

Transcript

Good morning, everybody. It's great to be here in San Francisco, the Bay Area. I used to live here during the nineties. I now live in, uh, Bangalore, and I'm gonna talk about this Aadhaar identity project in India. Let's look at India. India is a land of contrasts. One point two billion population, next only to China, but seventy percent of the peo... Read More

Key Insights

• Aadhaar is a common national identity platform intended to give every resident a portable identity and support online authentication. It addresses a system in which departmental paper credentials could be tied to one village, duplicated, or assigned to nonexistent people, limiting access and enabling subsidy leakage.
• Aadhaar enrollment is based on four demographic fields, ten fingerprints, two iris images, and a facial photograph. These records are assembled into an encrypted enrollment packet and processed in the Central Identities Data Repository before an identification number can be issued.
• Biometric deduplication is a compute-bound comparison against the entire enrolled database. The system checks a new resident’s fingerprints and irises against existing records one at a time because the biometric search cannot use a simple shortcut or conventional index.
• The Aadhaar number is completely random and contains no embedded personal information. It does not encode details such as the resident’s birthplace or date of birth, separating the identifier itself from the demographic and biometric information collected during enrollment.
• Fingerprint recognition works by comparing ridge minutiae, including islands, dots, and bifurcations, through their coordinates and orientations. Because agricultural work can wear down fingerprint ridges, fingerprints alone may not accurately serve every resident, making a second biometric mode necessary.
• Iris recognition provides a second biometric mode with a large amount of identifying information and an almost-zero false acceptance rate, according to the presentation. Combining iris data with fingerprints improves coverage for residents whose fingerprints cannot be captured accurately because their ridges are worn.
• Aadhaar’s enrollment ecosystem is distributed across public and private partners. State governments serve as registrars, private enrolling agencies operate field kits, certified operators capture residents’ information, and vendors and certification agencies supply and validate devices built to UID specifications.
• Aadhaar’s processing architecture is composed of independently scalable stages connected through RabbitMQ. Hadoop stores encrypted packets, MySQL handles relational data, MongoDB and Solr support horizontally scalable text search, and three biometric providers cross-check suspected duplicates while competing for dynamically allocated transactions.

Questions & Answers

Q: How does Aadhaar create a unique identity for a resident?

Aadhaar enrollment collects a resident’s name, address, gender, date of birth, ten fingerprints, two iris images, and a facial photograph. An encrypted enrollment packet is sent to the Central Identities Data Repository, where the biometric information is compared against the entire database. If the resident is found to be unique, the system issues a completely random Aadhaar number containing no embedded personal details.

Q: Why does Aadhaar use both fingerprints and iris images?

Aadhaar uses two biometric modes because fingerprints are not consistently reliable for every resident. Many people in India’s rural agricultural economy work with their hands, which can wear down fingerprint ridges and leave too few minutiae for accurate capture. Iris images provide substantial identifying information and an almost-zero false acceptance rate, according to the presentation, improving accuracy and reducing the risk of excluding farmers and other manual workers.

Q: How does Aadhaar biometric deduplication work?

Biometric deduplication determines whether an applicant is already present in the central repository. The system compares the applicant’s fingerprints and irises against the entire existing database, rather than relying on names, addresses, paper records, or a simple searchable index. This one-to-many comparison is compute-bound. Aadhaar also uses three biometric service providers, allowing suspected duplicates identified by one provider to be checked by the other two.

Q: What problem was Aadhaar designed to solve in India?

Aadhaar was designed to provide a common, portable identity for residents who often lacked standardized identification. Existing credentials could be paper-based, issued by separate government departments, and tied to the village where they originated. These systems also contained duplicate and ghost identities. Such weaknesses restricted migrants’ access to services and contributed to leakage from the forty billion dollars spent annually on subsidies and programs for poor residents.

Q: How does Aadhaar support online identity authentication?

Aadhaar authentication verifies whether a person is who they claim to be through an online request to the central system. A resident presents an Aadhaar number and may also provide a fingerprint at a service point such as a ration shop or bank. The request is checked against the authentication system, which returns a simple yes-or-no response rather than requiring the resident to depend on locally issued paper credentials.

Q: How was Aadhaar designed to enroll people across India?

Aadhaar relied on an ecosystem of partners to operate across India’s large, diverse population and uneven infrastructure. State governments became registrars, private enrolling agencies purchased field kits and conducted enrollment, and trained operators captured residents’ information. Training and certification organizations supported those operators, while device vendors supplied equipment built to UID specifications and certification agencies verified that the devices met the required standards.

Q: What technologies were used in Aadhaar’s central system?

Aadhaar’s central system was largely built with open-source technologies. Hadoop distributed file storage held encrypted enrollment packets with high availability. MySQL supported relational data, while MongoDB and Solr addressed horizontal scaling and text search. RabbitMQ connected independently scalable processing stages. The broader environment used Linux, Java, Spring AOP, commodity blade servers, and tools including Mule, Tomcat, Pig, Hive, and Zookeeper.

Q: Why did Aadhaar use three biometric service providers?

Aadhaar used three back-end biometric providers to improve deduplication accuracy and create performance-based competition. When one provider detected a possible duplicate, the relevant packets could be sent to the other two providers for additional checking. The system could also move transactions away from a weaker performer and toward a stronger one. Because providers earned money when Aadhaar numbers were generated, this dynamic allocation encouraged continued performance.

Summary & Key Takeaways

• Aadhaar was created as a common national identity platform for India’s residents, particularly those lacking portable, standardized identification. Enrollment collects four demographic fields, ten fingerprints, two iris images, and a facial photograph. The system checks each applicant against the full biometric database before issuing a completely random identification number.

• The enrollment challenge required cooperation among state governments, private enrolling agencies, trained and certified operators, device vendors, and certification organizations. This distributed ecosystem helped Aadhaar operate across urban and rural areas, multiple languages and scripts, differing literacy levels, and villages without broadband connections capable of transferring five-megabyte enrollment packets.

• Aadhaar’s central technology uses encrypted packets, Hadoop distributed storage, MySQL, MongoDB, Solr, RabbitMQ, Linux, Java, and independently scalable processing stages. Three competing biometric service providers perform deduplication, cross-check suspected duplicates, and receive dynamically reassigned workloads based on performance, improving accuracy while allowing capacity to expand through commodity blade servers.