Chapters

16 Wi-Fi Security: Provisioning and Recovery Evidence

iot
wi-fi
security-review

16.1 Start With the Decision

A device that cannot rotate a key will fail its next security review. Provisioning must cover updates, recovery, and safe removal from day one.

16.2 Route Overview

This is part 2 of 2. Review Wi-Fi Security: Access and Segmentation Controls for the preceding evidence.

16.3 Learning Objectives

  • Plan credential recovery, rotation, and device removal.
  • Review provisioning evidence for shared-key and per-device risks.

16.4 Chapter Roadmap

  • Recovery, Updates, And Removal
  • Worked Review: Small Building Sensors
  • Worked Review: Local Setup Device
  • Worked Review: Managed Facility Fleet
  • Knowledge Check: Shared Credential Risk
  • Knowledge Check: Provisioning Path
  • Match Evidence To Security Review Area
  • Order The Security Provisioning Review
  • Common Mistakes
  • Final Checklist
  • Authenticate, Then Derive Keys
  • The WPA2 Four-Way Handshake and Its Weakness
  • WPA3-SAE, PMF, and DPP Onboarding
  • Figure Review: Migration and Handshake State
  • Summary
  • Key Takeaway
  • What’s Next

16.5 Recovery, Updates, And Removal

Security design must survive normal operations.

Review:

  • firmware update path and rollback behavior
  • credential or certificate renewal path
  • device reset and reprovisioning flow
  • lost-device and stolen-device response
  • decommissioning path before resale, disposal, or transfer
  • support access limits and audit trail
  • monitoring and alert ownership

Security fails when a design is secure only on the first day. The review should include maintenance and end-of-life behavior before approval.

16.6 Worked Review: Small Building Sensors

Scenario:

  • A building sensor fleet uses Wi-Fi for periodic telemetry and occasional maintenance commands.

Review route:

  • Authentication: decide whether shared-password mode is acceptable or whether per-device identity is needed.
  • Provisioning: check installer flow, setup timeout, reset, and support workflow.
  • Segmentation: allow telemetry and update paths, block unrelated internal paths.
  • Recovery: define lost-device removal, credential rotation, and firmware update owner.
  • Decision: accept only if the operational team can maintain the chosen security model.

16.7 Worked Review: Local Setup Device

Scenario:

  • A device creates a temporary setup path during installation.

Review route:

  • Authentication: require physical-possession proof or a bounded pairing step.
  • Provisioning: close setup mode after success, timeout, reset, or failed attempts.
  • Credential handling: keep secrets out of logs, UI screenshots, source code, and support messages.
  • Recovery: document how a user restarts setup without exposing the previous credential.
  • Decision: accept only if setup convenience does not leave a permanent open management path.

16.8 Worked Review: Managed Facility Fleet

Scenario:

  • A facility operates multiple device classes with different owners and service paths.

Review route:

  • Authentication: prefer per-device identity when fleet operations can renew and revoke credentials.
  • Provisioning: use managed enrollment and record ownership for identity lifecycle.
  • Segmentation: separate device classes by service need, not just by convenience.
  • Monitoring: alert on unexpected destinations, repeated join failures, and unsupported security modes.
  • Decision: accept only if identity lifecycle, network policy, and support ownership are documented.

16.9 Knowledge Check: Shared Credential Risk

16.10 Knowledge Check: Provisioning Path

16.11 Match Evidence To Security Review Area

16.12 Order The Security Provisioning Review

16.13 Common Mistakes

Treating onboarding as separate from security:

  • Problem: setup mode works, but it leaves credentials or management access exposed.
  • Repair: review setup entry, exit, timeout, reset, and credential handling as part of the security boundary.

Using one shared credential without a recovery plan:

  • Problem: one leaked or lost device can force a disruptive fleet-wide credential change.
  • Repair: document rotation, revocation, segmentation, owner, and whether per-device identity is needed.

Assuming the newest mode is enough:

  • Problem: the label on the AP does not prove client support, provisioning safety, update behavior, or segmentation.
  • Repair: review the full device lifecycle and measured join behavior.

Leaving support paths unbounded:

  • Problem: a support feature becomes a permanent management opening.
  • Repair: define who can enable it, when it closes, what it logs, and what forces a retest.

Forgetting decommissioning:

  • Problem: devices leave service with credentials or owner binding still present.
  • Repair: include transfer, wipe, removal, and identity retirement in the acceptance checklist.

16.14 Final Checklist

Before accepting a Wi-Fi security and provisioning design, confirm that the review includes:

  • device class, data sensitivity, command risk, and owner
  • selected authentication mode and blocked fallback modes
  • provisioning path and physical-possession or pairing proof
  • setup timeout, reset, failure, and support behavior
  • credential storage, rotation, revocation, and removal plan
  • network segment and allowed service rules
  • update, recovery, and decommissioning path
  • monitoring signal and response owner
  • accepted decision and rejected alternatives
  • retest triggers for firmware, AP profile, credential policy, installer flow, service path, or ownership changes

16.15 Authenticate, Then Derive Keys

Wi-Fi security answers two questions: is this device allowed on the network, and how do the two ends agree on encryption keys? For home and IoT networks the dominant schemes are WPA2-Personal and its successor WPA3-Personal. Both start from a shared password, but they turn it into keys very differently — and that difference decides how resistant the network is to attack.

A separate but linked problem is provisioning: how a screenless IoT device gets the Wi-Fi credentials in the first place. Typing a passphrase into a sensor with no keyboard is impossible, so the industry built dedicated onboarding methods.

The review should keep those jobs separate. A strong key-agreement protocol does not prove that setup mode closes, credentials are stored safely, one lost device can be removed, or the device is isolated after it joins. Likewise, a convenient provisioning app does not prove that the final network session uses the intended authentication mode.

For an IoT fleet, the first evidence record should name the authentication mode, fallback modes that are blocked, provisioning entry and exit rules, credential owner, storage location, rotation or revocation path, segmentation boundary, update path, monitoring signal, and retest trigger. That record turns “uses WPA3” or “uses a QR code” into an auditable lifecycle decision.

A simple acceptance test can stay lightweight: join a representative device, confirm the negotiated security mode on the access point, confirm the provisioning path cannot be restarted without authorization, then remove the device and verify that it cannot silently rejoin from stale credentials. Keep screenshots or controller exports with the record. Those checks expose the failure modes that matter most in long-lived field deployments.

Two jobs: a key-agreement scheme (WPA2 or WPA3) secures the link, and a provisioning method gets credentials onto a device with no screen or keyboard.

16.15.1 Overview Knowledge Check

16.16 The WPA2 Four-Way Handshake and Its Weakness

WPA2-Personal turns the passphrase into a Pairwise Master Key (PMK), then runs the four-way handshake to derive the per-session Pairwise Transient Key (PTK) that actually encrypts traffic:

  1. AP → station: a random ANonce.
  2. Station → AP: its own SNonce plus a MIC. Both sides now derive the PTK from PMK + both nonces + both MAC addresses.
  3. AP → station: the group key (GTK) and a MIC, confirming keys.
  4. Station → AP: acknowledgment; encryption begins.

Use Figure 16.1 to trace which values cross the air and which keys are derived locally. That distinction is necessary before judging what a captured handshake reveals and why passphrase quality still matters.

WPA2 four-way handshake showing supplicant and authenticator exchanging ANonce, SNonce, MIC, GTK, acknowledgment, and deriving PTK and GTK session keys.
Figure 16.1: Use the handshake diagram to separate proof of session-key establishment from the broader fleet questions: password strength, shared-secret exposure, rotation, segmentation, and device removal.

Read Figure 16.1 from the AP’s ANonce to the station’s SNonce and MIC, then through the AP’s protected GTK delivery and the final acknowledgement. The passphrase itself is not transmitted; both peers derive the PTK from the PMK, nonces, and addresses, while MICs prove possession. A capture therefore supplies material for offline guesses: an attacker can derive candidate keys and test the MIC without contacting the AP. This connects session establishment to the chapter’s fleet controls for passphrase entropy, rotation, segmentation, and device removal.

Worked example. An attacker sniffs a WPA2 network’s four-way handshake, then runs a wordlist offline against it. If the passphrase is a common phrase, it is recovered in minutes; a long random passphrase resists it. This offline-guessing exposure is exactly what WPA3 was designed to close.

In a defensive review, the point is not to reproduce an attack. It is to decide whether a shared passphrase is acceptable for this device class. Ask who knows the passphrase, how it is provisioned, whether it ever appears in logs or support messages, how one lost device is handled, and whether the network segment limits the damage if one credential leaks.

For WPA2-Personal fleets that cannot move immediately, the compensating controls should be explicit: unique SSIDs or VLANs for device classes, a high-entropy passphrase controlled by operations rather than end users, documented rotation triggers, disabled WPS, and monitoring for repeated authentication failures or unexpected reassociation after removal.

16.16.1 KRACK: The Key-Reinstallation State Failure

The four-way handshake must tolerate a lost final acknowledgement. If message 4 does not reach the AP, the AP retransmits message 3. Retransmission itself is correct. The failure appears when a client treats a valid retransmission as a command to install the same pairwise key again and resets state that must never move backwards.

Trace the vulnerable sequence:

  1. The AP and client derive a PTK from the PMK, both nonces, and both MAC addresses.
  2. Message 3 tells the client to install that key and carries group-key information and a replay counter.
  3. A man-in-the-middle delays or suppresses message 4, causing the AP to resend message 3.
  4. A vulnerable client accepts the authentic retransmission but reinstalls the already-used PTK.
  5. Reinstallation resets a transmit packet number/nonce and may reset receive replay state. Some affected Linux/Android behavior additionally installed an all-zero key; that was a particularly severe implementation case, not the behavior of every client.
  6. Encryption now reuses a nonce with the same key. For stream-like counter encryption, C=PKS(K,nonce)C=P\oplus KS(K,nonce), so nonce reuse repeats keystream and can expose relationships between plaintexts, enable replay, or permit forgery depending on the negotiated cipher and traffic.

The repair is a state-machine invariant: after a key is installed, a retransmitted message may be acknowledged but must not reinstall that key or reset its nonce/replay counters. Test duplicate, reordered, and lost handshake messages against that invariant. WPA2 devices with patched supplicants can be safe from this flaw; migrating protocol labels without updating vulnerable implementations is not a fix.

WPA3 strengthens the surrounding design but its features solve different problems. SAE replaces PSK establishment with a password-authenticated exchange that resists passive offline guessing and provides forward secrecy. Enhanced Open/OWE encrypts traffic on otherwise open networks; it does not authenticate the access point and does not “replace PSK” on a protected personal network. DPP/Easy Connect gives limited-UI devices a public-key onboarding path. WPA3-Enterprise’s 192-bit security mode is an optional high-security enterprise profile, not the default for every IoT client. Record the negotiated suite, Protected Management Frames requirement, transition-mode fallback, key-installation regression test, and per-device onboarding result instead of recording only “WPA3 capable.”

16.16.2 Practitioner Knowledge Check

16.17 WPA3-SAE, PMF, and DPP Onboarding

WPA3-Personal replaces the PMK-from-passphrase step with SAE (Simultaneous Authentication of Equals), a password-authenticated key exchange (the “Dragonfly” handshake). Its crucial property: an attacker cannot test password guesses offline — each guess requires a fresh online exchange with the AP, so brute force is throttled to network speed, and SAE also gives forward secrecy (a later password compromise cannot decrypt past traffic). WPA3 additionally mandates Protected Management Frames (802.11w), which authenticate deauthentication and disassociation frames to block the deauth-flood attacks that plague WPA2.

For onboarding screenless devices, Wi-Fi Easy Connect (DPP) uses public-key cryptography: a configurator (say a phone app) reads the device’s bootstrapping key — often from a printed QR code — and then securely hands the device its network credentials. No passphrase is ever typed into the device, and the network PSK is never exposed to it in the clear. (Simpler schemes exist too: a device may host a temporary SoftAP or use BLE to receive credentials from an app.)

Worked example. A new smart sensor ships with a QR code on its label. During setup the installer scans it with the vendor app (the configurator); DPP uses the encoded public key to establish a secure channel and provision the sensor onto the WPA3 network — no keyboard, no exposed PSK, and management frames protected by PMF against a deauth attacker trying to disrupt the onboarding.

The under-the-hood acceptance evidence is the negotiated behavior, not the label on the box. Record whether the AP and client actually use WPA3-SAE, whether transition mode or WPA2 fallback is allowed, whether PMF is required, how DPP bootstrapping keys are protected, and how the device leaves setup mode after success, timeout, or failure. Also record what invalidates the approval: AP profile change, firmware change, installer app change, QR label process change, credential rotation, ownership transfer, or service-path change.

16.17.1 Under-the-Hood Knowledge Check

16.18 Figure Review: Migration and Handshake State

Legacy Wi-Fi security is easiest to reject when its failure mechanisms are traced through the packet-protection sequence rather than reduced to an age label.

Side-by-side WEP and TKIP mechanism sequence showing WEP secret key plus visible IV feeding RC4, repeated-IV and CRC weaknesses, then TKIP per-packet key mixing, replay checks, weak Michael MIC, and migration to AES-CCMP or WPA3.
Figure 16.2: WEP failure and TKIP migration sequence.

In Figure 16.2, visible IV · 24 bits and RC4 keystream lead to reuse and linear-integrity failures; TKIP · TEMPORARY BRIDGE adds mixing and replay checks but retains RC4 and the weak MIC “Michael”, so the route ends at AES-CCMP or WPA3.

Modern protocol review must also distinguish a sound key exchange from an unsafe implementation state transition.

WPA2 message 3 must install the PTK and reset nonce once; retransmitted M3 is for ACK recovery. The WPA3 map links SAE, OWE, DPP, enterprise strength and forward secrecy to distinct threat boundaries.
Figure 16.3: WPA2 KRACK state and WPA3 mechanism map.

Figure 16.3 marks M3 · install PTK + reset nonce ONCE and the unsafe retransmitted-M3 reinstallation that enabled KRACK; the WPA3 MECHANISM MAP then assigns SAE, OWE, DPP, and the 192-bit suite to different threat boundaries.

16.19 Summary

Wi-Fi security and provisioning review is a lifecycle decision. The design must be secure when the device joins, fails setup, receives updates, changes owner, loses credentials, is removed from service, or needs support.

The final output should be a bounded decision record: authentication mode, provisioning path, credential owner, storage and revocation behavior, segmentation boundary, maintenance path, monitoring signal, and retest triggers.

16.20 Key Takeaway

Wi-Fi Security and Provisioning should align Wi-Fi onboarding, credentials, encryption, rotation, recovery, and support visibility with deployment evidence.

16.21 What’s Next

Use Wi-Fi Deployment Planning when the security decision must be validated across a site rollout.

Use Wi-Fi 6 Density Review when the AP profile or client support decision affects dense-device behavior.

Use Wi-Fi Implementation: Comprehensive Lab when the reviewed security and provisioning plan is ready for release validation.

Use Wi-Fi Standards Index when the team needs to choose the right standards route before making a security decision.

16.22 Continue Your Route

This final part closes the route from Recovery, Updates, And Removal through What’s Next. Return to Wi-Fi Security: Access and Segmentation Controls or continue from the wifi-mobile module index.