Parking Access Control: ANPR vs Cards vs Mobile — Cost, ROI, Migration Path

Most Swiss commercial buildings still run on access-control hardware specified in 2013–2016. The technology has moved twice since then. This article compares the three dominant approaches — ANPR cameras, RFID/proximity cards, and mobile credentials — on CAPEX, OPEX, throughput, dispute volume, and integration risk, with a phased migration plan that doesn’t require ripping out the existing barriers.

What “access control” actually delivers

In a commercial parking context, access control is the layer that decides three things in real time:

• Who is at the barrier right now — matching the inbound vehicle to a known credential (subscriber, hourly visitor, delivery vehicle, blacklisted plate).
• What rate applies — the rate engine takes the credential class, the time, and the dynamic-pricing rule set (see playbook) and produces a billable rate or a no-charge pass.
• Whether to open the barrier — the decision flows back to the gate within the timing budget the operator has set (typically 600–1500 ms from credential read to barrier action).

The three dominant technologies all achieve those three outcomes, but with very different trade-offs on operator OPEX, customer experience, and integration complexity. We’ll walk each one in turn, then put them side by side.

Technology 1: ANPR (Automatic Number Plate Recognition)

ANPR uses a fixed camera at the barrier reading the licence plate, OCR-ing it into a plate string, and looking the string up against the credential database. No physical token is needed — the vehicle’s plate is the credential.

Hardware

• 1 camera per lane, mounted ~3 m before the barrier with IR illumination for night reads.
• Edge compute box running the recognition model. Modern systems do OCR on-device with <200 ms latency; legacy systems shipped the image to a cloud endpoint, adding 400–800 ms.
• Optional second camera at the exit for billing reconciliation and unrecognised-plate flagging.

Operator economics

Hardware CAPEX: CHF 3,500–6,500 per lane for Swiss-spec installations including conduit work and integration. OPEX: low (no tokens to issue, lose, or replace; software subscription typically CHF 80–180/month per site).

Customer experience

The best UX of the three technologies. Customer drives up, barrier opens within 1–2 seconds, no phone fiddling, no card to fumble for. Same on exit. Throughput: typically 12–18 vehicles per minute per lane, vs. 6–10 for card systems. Critical for office buildings with 100+ vehicles arriving in a 30-minute morning window.

Failure modes

• Plate occluded (dirty plate, bike rack mounted in front, snow build-up). Modern systems handle this by falling back to a manual barrier-open with an audible chime; 0.3–0.8 % of reads on a healthy installation.
• Plate change (new car, dealer plate during temporary use). Subscribers need to register additional plates — typically self-service via the operator’s portal.
• Privacy regulation — ANPR is regulated under EU GDPR and Swiss FADP as biometric/personal data when plate-to-person mapping is stored. Audit-quality vendors anonymise after billing.

Best fit

Office, mixed-use, and large residential with predictable plate populations. Hospitals and event venues where throughput at peak matters more than anything else.

Technology 2: RFID / Proximity Cards

The incumbent. Each authorised user carries an RFID card (typically 125 kHz prox or 13.56 MHz MIFARE), tapping or waving it at a reader 30–100 cm from the driver’s seat. Reader sends the card ID to the access controller, controller checks the credential database, barrier opens.

Hardware

• 1 long-range reader per lane, mounted at driver-window height.
• Access controller (often a Vanderbilt, Bosch, or Salto unit) wired to barrier I/O.
• Card stock — physical cards issued per user.

Operator economics

Hardware CAPEX: CHF 2,000–4,200 per lane (cheaper than ANPR upfront). OPEX is where it hurts:

• Card stock: CHF 8–25 per card, plus printing and personalisation.
• Replacement rate: 8–15 % of card population per year (lost, broken, demagnetised). For a 200-spot building that’s 16–30 cards to re-issue annually, each generating an admin ticket.
• Office overhead: 0.3–0.6 admin hours per replacement card.
• Lock-in: once you’ve chosen a card format (HID prox vs Mifare DESFire vs Salto SVN) you’re tied to that vendor’s ecosystem. Migrations are CHF 60–180 per spot.

Customer experience

Adequate but not great. The card has to be within 30–100 cm of the reader; busy users juggling kids, coffee, and a card on a lanyard often miss the read window. Throughput drops to 6–10 vehicles per minute per lane. New subscribers wait days for their card to arrive.

Failure modes

• Lost card — until reported, the holder is locked out, and an unauthorised holder may have access. Mitigated by quick deactivation (most systems support same-day) but the operational friction is real.
• Cloned card — older 125 kHz prox formats are trivially cloneable. MIFARE DESFire raises the bar significantly; Salto SVN is currently the most secure mainstream Swiss option.
• Battery-free readers in cold environments — the read range can drop in winter; not a meaningful problem in indoor garages.

Best fit

Small properties with stable, low-churn subscriber bases (residential, small offices). Where ANPR can’t be installed because of camera angle constraints (very tight entry, mixed pedestrian-vehicle traffic).

Technology 3: Mobile credentials (smartphone)

The current generation. Each authorised user installs an app (or scans a wallet-style pass into Apple/Google Wallet), and the phone acts as the credential. The barrier reader uses Bluetooth Low Energy (BLE) or NFC to detect the phone’s credential token within range.

Hardware

• BLE/NFC reader per lane — cheaper than ANPR cameras, comparable to high-end RFID readers.
• Operator-side app or wallet pass infrastructure.
• Optional integration with payment provider for hourly visitor flow.

Operator economics

Hardware CAPEX: CHF 2,500–4,800 per lane. OPEX low (no physical tokens; operating system covers credential lifecycle). The mobile-credential SaaS layer typically runs CHF 120–280/month per site for a small property.

Customer experience

Mixed. For tech-comfortable users (most office subscribers): excellent — no card to carry, phone in pocket, barrier just opens. For older residents or visitors without the app installed: worse than RFID — they have to install something on their phone before they can park.

Throughput: 10–14 vehicles per minute per lane — better than RFID, slightly worse than ANPR because BLE handshake takes ~600–1200 ms in cold-cache conditions (first vehicle of the morning).

Failure modes

• Phone battery dead — user can’t enter. Mitigated by a fallback PIN at the barrier or a one-time SMS code.
• App not installed — new visitors need to provision before they can park. Operator portals should auto-SMS the install link on first reservation.
• OS update breaking BLE behaviour — happens once every 12–24 months on iOS. Vendor needs to ship a hotfix within ~48 hours; mature vendors do this routinely.
• Bluetooth privacy concerns — iOS 17+ and modern Android randomise BLE MAC addresses; well-built systems use rotating credential tokens, not raw MAC.

Best fit

Hotels (guest already has the app via the booking flow), corporate offices with smartphone-issued workforce, residential buildings where the operator already runs a tenant app.

Side-by-side comparison

Migration path: from legacy to modern without ripping out the barriers

The good news for Swiss operators with 10-year-old card systems: the barriers themselves rarely need to be replaced. Modern barriers from FAAC, BFT, and Magnetic use the same I/O protocol that 2015-era barriers used. The migration is really a software + reader-head swap.

Phased plan we’ve run on 23 buildings since 2024:

Realistic outcomes for a 100-spot Swiss office running this migration:

• Year-1 net cost: ~CHF 12,000 (hardware + integration + the labour involved in enrolment campaigns).
• Year-1 savings: CHF 8,500–14,000 (eliminated card stock, fewer admin tickets, no replacement cards, lower dispute volume).
• Payback: 12–18 months on the hardware alone, faster if combined with the dynamic-pricing rollout (which the new access controller enables).

Where the dynamic-pricing playbook plugs in

None of this matters if the access layer doesn’t talk to the pricing engine. The four dynamic-pricing mechanisms (time-of-day, occupancy, event, weather) all depend on the access controller being able to:

• Identify the credential class within the timing budget (~600 ms),
• Query the rate engine in parallel with credential validation,
• Display the current rate at the entry barrier (or quote it via the reservation channel) before the customer commits, and
• Push the realised rate back to billing for reconciliation against the credential.

Modern ANPR-based stacks do all four out of the box. Most legacy card stacks do only the first; the dynamic-pricing logic has to be retrofitted via middleware, which adds 6–14 weeks to any pricing rollout. That’s the strongest single argument for sequencing the access-control modernisation before the pricing rollout, not the other way around.