Biometric Sensor and Security Solutions

A biometric check compares a live trait with a stored template. You will explain enrolment, templates and matching, and compare common traits, so you can describe what a biometric check actually does.

• Types: fingerprint, face, iris, voice, vein pattern.
• Enrol once, then match each time; a score decides pass or fail.
• Templates are numbers, not photos, but still personal data.
• Match decisions use a threshold you can move.
• Learn the difference between enrolment, which stores a template, and matching, which compares a new sample with it; each step has its own error sources.
• List trait options such as fingerprint, face, iris and voice, and note for each one how easily it can be copied, how it changes with age and what hardware it needs.
• Learn the difference between identification (who are you?) and verification (are you who you say?).
• Compare how often each trait changes: faces and voices change more than fingerprints.

Activity. Compare three biometric types on ease, cost and risk.

Standard to meet. Complete the table with at least four points each.

Common mistake. Believing a match score is a yes or no fact.

Check yourself. What threshold did you choose and why?

Ask whether a simpler lock solves the problem. You will state a security need in one sentence and test whether cards or codes already meet it before considering biometrics.

• Define who must enter, how often and what is at risk.
• List alternatives: key, card, code, guard.
• Biometrics suit high-risk or high-volume places, not every door.
• Count cost over three years, not just purchase price.
• Write the need as a sentence with who, where and what is at stake, such as 'staff opening the storeroom, 20 people, low risk', before looking at products.
• Compare simpler options such as cards or codes against biometrics on cost, speed, loss and misuse, and keep the simple one if it meets the need.
• Ask users what would make them comfortable; their answer shapes the privacy notice.
• Check whether a mobile phone unlock or a smart card already meets the need.

Activity. Write a need statement and compare two tools.

Standard to meet. Write a specific need statement and show the cost over 3 years.

Common mistake. Buying a reader before stating the need.

Check yourself. What is your need in one sentence?

Trust is the product. Collect the least and say so. You will draft a consent note and a retention rule that collect the least data, protecting both users and the organisation's credibility.

• Written, informed consent; no pressure.
• Offer an equal non-biometric way in.
• Store templates encrypted; never share or sell.
• Set a retention time and a deletion route.
• Write a plain consent note saying what is collected, why, how long it is kept, who sees it and how a person can say no or ask for deletion.
• Store templates rather than raw images where possible, encrypt them, and set a deletion date so data does not sit around after its purpose ends.
• Keep consent forms apart from the templates so one leak does not link name to data.
• Name a person responsible for deletion requests.

Activity. Write a one-page privacy notice and consent form.

Standard to meet. Cover purpose, storage, deletion and alternative in the notice.

Common mistake. Hiding the alternative path in small print.

Check yourself. Where is the alternative described?

Install carefully; protect the device. You will run a small, well-recorded trial with volunteers under varied conditions, giving you evidence rather than a supplier's claims.

• Fit at right height and light; keep sensors clean.
• Change default codes; limit admin rights.
• Enrol volunteers with two samples each.
• Keep a log of faults.
• Enrol 20 to 30 volunteers, record conditions such as light and dry or wet hands, and run at least 100 genuine attempts in total, split across your volunteers.
• Fix the placement of the sensor at a comfortable height and test it with left and right hands and with people of different heights.
• Take photos of the installation for the log, avoiding any faces.
• Test with gloves, dust and bright sunlight, which are the usual failures.

Activity. Install or simulate the reader and enrol volunteers.

Standard to meet. Enrol all volunteers and start the log.

Common mistake. Skipping the second enrolment sample.

Check yourself. How many samples per person?

Accuracy is measured, not promised. You will calculate false accept and false reject rates from trial counts and tune a threshold to the error that costs more.

• False accept: wrong person let in. False reject: right person refused.
• Rates = errors divided by attempts.
• Moving the threshold trades one error for the other.
• Test with different people, lighting and wet or dry hands.
• Compute the false accept rate, where a wrong person is let in, and the false reject rate, where the right person is turned away, from your trial counts.
• Move the threshold in small steps and plot both rates; the point where they meet shows the trade-off, and your choice depends on which error costs more.
• Make sure at least 100 attempts include different people, not one person many times.
• Record the threshold used for each test run.

Activity. Run 100 or more attempts and compute both rates.

Standard to meet. Calculate and explain both rates.

Common mistake. Testing only with easy cases.

Check yourself. What conditions did you test?

Attackers go after the weak link, not the reader. You will map the whole chain from sensor to door controller and list a likely attack at each link, adding liveness checks where justified.

• Spoofing: photos, copies; ask about liveness checks (tests that the trait comes from a living person).
• Protect cables, power and the network.
• Log entries and review them weekly.
• Plan for lost power and for a person who cannot enrol.
• Map the chain from sensor to template storage to the door controller, and list one possible attack at each step, such as a fake finger or intercepted data.
• Add a liveness check or a second factor where the risk justifies it, and keep logs of failed attempts so repeated tries trigger an alert.
• Keep the network for the readers separate from the guest Wi-Fi.
• Test the door release on power loss with the safety officer present.

Activity. Write a threat table with five items.

Standard to meet. Give each threat a control.

Common mistake. Forgetting power loss.

Check yourself. What happens to your door in a power cut?

Systems must work for everyone. You will compare error rates across groups and design a respectful non-biometric alternative, so the system serves everyone without penalty.

• Some people cannot use a trait: injury, age, work wear.
• Check results across groups for gaps.
• Offer card or code and keep it equally fast.
• Handle complaints quickly.
• Compare error rates across groups such as age bands and skin tones in your trial, and flag any group whose rate is more than double the average.
• Provide a non-biometric route, such as a card and a staff check, and make sure using it carries no penalty or delay beyond a few minutes.
• Review complaints and add each to the fairness log.
• Measure the time to enter through each route; the alternative should be about as quick.

Activity. Compare results across at least two groups.

Standard to meet. Name the gap and propose a remedy.

Common mistake. Ignoring group differences.

Check yourself. Which group had the highest rejection rate?

Say yes, no or not yet with evidence. You will write a one-page recommendation separating measured from estimated figures and schedule a re-check, so your yes, no or not yet stays honest.

• Compare error rates, cost and privacy risk with the need.
• Write a one-page recommendation.
• Plan a six-month review.
• Delete trial data and record when.
• Summarise on one page the need, trial size, error rates, privacy steps, cost and your recommendation, and mark which numbers are measured and which are estimates.
• Schedule a review at six months to repeat the error checks, because wear, software updates and new users change results over time.
• Plan a re-test every six months, because hardware ages.
• Write down what would make you stop using biometrics.

Activity. Write the recommendation and 90-day plan.

Standard to meet. Back the decision with numbers and record the deletion.

Common mistake. Keeping trial data "just in case".

Check yourself. When will your trial data be deleted?

An employee does not want a fingerprint reader.

Question: What must you do?

Model answer: Offer an equal alternative at once, with no penalty, and record the choice.

A tester opens the face reader using a photo.

Question: What does this show?

Model answer: The system lacks a liveness check; add it or choose another method and record the weakness.

On the slide: L1 Basic · Basics of access

Notes: Explain how a lock, a key card, a PIN and a fingerprint each prove who you are, and what each can get wrong. Notes: look at the doors in your building and list how each is opened. Standard: a one-page list a visitor could follow.

On the slide: L2 Awareness · Awareness

Notes: Describe what a biometric is, why it cannot be changed if leaked, and when its use feels fair or unfair to the person using it. Notes: discuss two cases, a school gate and a gym, and write one rule you would set for each. Standard: your rules name consent, a choice without biometrics and who can see the data.

On the slide: L3 Foundation · Foundation

Notes: Compare a card, a PIN and a biometric reader for one need, such as a staff door, on cost, error, privacy and fallback. Notes: paper exercise with a scoring table and a written reason for the choice. Standard: scores come from stated numbers, and the fallback is described.

On the slide: L4 Practitioner · Practitioner

Notes: Run a small trial of an entry solution with a simulator or a card system, count false accepts and false rejects over at least 50 attempts, and test the no-biometric route. Notes: simulator allowed; no real fingerprints or faces stored. Standard: error rates are calculated and the alternative route works.

On the slide: L5 Specialist · Specialist

Notes: Design and trial a small access-control solution with a consent process, measured errors and a no-biometric alternative, for a site of 20 people. Notes: write the notice, the retention period and the spoofing test you ran. Standard: error rates reported, consent form tested on a reader, alternative route in daily use.

On the slide: L6 Advanced · Advanced

Notes: Design access control for several sites, with a standards check, an audit trail and an independent test. Notes: add a testing lab report and a data-protection review for each site. Standard: the audit trail answers who entered, when and on whose approval, and the review is dated.

On the slide: L7 Super expert · Super expert

Notes: Set the enrolment, template and fallback standard for biometric systems and train the installers who apply it. Notes: publish consent and spoofing-test templates and review false-match and false-reject figures each year.