Skip to content

Back to the catalog

Fast and Secure Simultaneous Authentication of Equals for WPA3
arxiv.orgpaper

Fast and Secure Simultaneous Authentication of Equals for WPA3

This paper addresses CPU-exhaustion Denial-of-Service risk in WPA3's Simultaneous Authentication of Equals (SAE) handshake, whose Password Element derivation is computationally expensive for the Access Point. The authors propose an asymmetric protocol modification that shifts the costly iterative PE-discovery work to the client, adds a slow-path key derivation step (using PBKDF2 or Argon2) to further throttle brute-force attempts, and introduces a ticket-based fast re-authentication path for known devices — all aimed at keeping APs available under adversarial load without weakening SAE's dictionary-attack resistance.

TL;DR A proposed asymmetric-cost, ticket-augmented redesign of WPA3's SAE handshake to stop CPU-exhaustion DoS attacks on access points.
Document

What it is

An academic proposal (arXiv:2609.31519v1, Ferreira/Zúquete/Gomes, University of Aveiro, Sep 2026) for a modified WPA3-SAE handshake that removes the AP's CPU-exhaustion DoS exposure. It is implemented — the modified hostapd v2.10 code is public at https://github.com/joaopferreira01/fast-sae-wpa3. The design splits authentication into two paths: a Resource Intensive Initial Exchange (RIIE) and a Stateless Fast Track Resumption (SFTR).

How it works

The summary states that PE-discovery moves to the client; the page gives the mechanism that makes single-step AP verification possible. Instead of both peers iterating Hunting-and-Pecking, SAE_ctr is derived deterministically as SAE_ctr = HMAC_pwd_hash(s), where pwd_hash = H(PSK). The Supplicant iterates (incrementing p, recomputing s) until it finds a valid curve point; the AP recomputes the same SAE_ctr from the received s and its cached pwd_hash in one step.

This introduces a new weakness the authors acknowledge: because the AP no longer iterates, an eavesdropper can test PSK guesses directly against the observed s, collapsing the offline-attack cost. That is precisely why H must be a slow KDF (PBKDF2/Argon2) — the slow path is not just throttling, it replaces the brute-force resistance the iteration loop used to provide. Iteration count and salt are broadcast in a vendor-specific Beacon/Probe IE (16-byte salt + 2-byte iteration count; the captured example shows 4096 iterations).

Ticket resumption (SFTR) is signalled by overloading s in the Commit frame, which normally must exceed 1:

  • Supplicant sends s = 0 plus the ticket in place of E.
  • AP accepts → replies s = 0 with random E.
  • AP rejects → replies s = 1 and terminates.

Tickets carry a sequential TID and a future random PMK, encrypted under a locally-held Secret Ticket Encapsulation Key (STEK) using Encrypt-Then-MAC. The AP stays stateless; replay is blocked by an IPSec-style sliding-window check on TIDs, and STEK refresh revokes all outstanding tickets (relevant after a PSK change meant to evict users).

flowchart TD
    A[SAE Commit received] --> B{s value}
    B -->|s=0 + ticket| C[Validate STEK-MAC, recover PMK']
    B -->|s>1| D[RIIE: recompute SAE_ctr from s, single-step PE verify]
    C -->|valid| E[Reply s=0, proceed to 4-Way HS]
    C -->|invalid| F[Reply s=1, terminate]
    D --> G[Issue new STEK-encrypted ticket]

Caveats

This is a proposal with experimental results asserted but no numbers in the fetched text — no benchmark figures, DoS thresholds, or comparison metrics are quoted beyond the qualitative claim that per-request AP load becomes constant. The security analysis section is truncated. Treat the s-overloading and STEK design as a research design to evaluate, not a standardised or independently reviewed mechanism; the shift of dictionary-attack resistance onto the KDF means the chosen iteration count (4096 in the demo) is now the load-bearing parameter.

Organisations named here

  • Password Element
  • Simultaneous Authentication of Equals

Technologies named here

  • WPA3
  • Argon2
  • PBKDF2
  • SAE

Techniques named here

  • CPU exhaustion Denial-of-Service attack
  • Offline dictionary attack