Skip to content
STIMSMITH

AES Key Schedule

Concept WIKI v1 · 7/6/2026

The AES Key Schedule is the key-expansion procedure that derives round keys from the AES master key and is widely modeled and targeted in fault-based and side-channel analyses of AES-128 implementations.

AES Key Schedule

Overview

The AES key schedule is the component of the Advanced Encryption Standard (AES) that expands a master key into a sequence of round keys used by the cipher's data-processing rounds. In AES-128, the schedule generates 11 round keys (including the initial key) from the 128-bit master key, and the same expansion structure is reused (with different round counts) for AES-192 and AES-256.

Role in Fault-Based Cryptanalysis

The key schedule is an attractive target for Differential Fault Analysis (DFA) because faults injected during key expansion can propagate to multiple downstream round keys, amplifying an attacker's control over the cipher with relatively few injected faults.

Use Case III in µARCHIFI

In the µARCHIFI evaluation, Use Case III instantiates a concrete analysis of an AES-128 key schedule:

  • Program: KeySchedule (AES), 1983 instructions, compiled with -Os.
  • Hardware: Baseline Ibex RISC-V core with no countermeasures.
  • Attacker goal: φIII := (9th Round key byte = 0) — zeroing a byte in the penultimate round key.
  • Property name in the case-study table: "Set expanded key to 0".
  • Fault model: Reset faults injected into combinational logic in the EX stage.
  • Fault injection budget: N = 2 (verified unreachable; a more powerful attacker reaches the goal with four injections).
  • Verification result: Property φ unreachable under the considered fault model.
  • Implementation reference: The key-schedule implementation analyzed is kokke's Tiny AES.

Differential Fault Analysis on the Key Schedule

The literature cited in µARCHIFI explicitly identifies the key schedule as a DFA target:

  • Takahashi, Fukunaga, and Yamakoshi — DFA Mechanism on the AES Key Schedule (FDTC 2007).
  • Ali and Mukhopadhyay — A Differential Fault Analysis on AES Key Schedule Using Single Fault (FDTC 2011).

Both works demonstrate that fault injections localized to the key-expansion procedure can yield exploitable biases, often with fewer faults than are required to attack the data rounds directly.

Cold-Boot / Neural Reconstruction

The structure of the AES key schedule has also been exploited in cold-boot key-recovery attacks. The key-expansion process introduces fixed redundancies across round keys, enabling recovery of the master key from partially corrupted memory. Recent work formalizes the AES key scheduling as a computational graph and implements it as a neural message-passing network, outperforming prior cold-boot attack methods.

Related Concepts

  • Differential Fault Analysis (DFA) — cryptanalytic technique that uses faulty ciphertexts/round keys to recover secrets; key-schedule-targeted variants are a specialization of DFA.
  • Tiny AES — compact C reference implementation of AES (kokke), including the key schedule, used as the program-under-analysis in the µARCHIFI AES case study.
  • µARCHIFI — formal verification tool for microarchitectural fault injections; contains the AES key-schedule case study described above.

CITATIONS

4 sources
4 citations
[1] The AES key schedule is targeted by Differential Fault Analysis (Takahashi et al. 2007; Ali and Mukhopadhyay 2011). µArchiFI: Formal Modeling and Verification Strategies for Microarchitectural Fault Injections
[2] Use Case III in µARCHIFI analyzes a 128-bit AES KeySchedule program compiled with -Os on the Ibex core, with attacker goal φIII := (9th Round key byte = 0) under combinational reset faults in the EX stage and a budget of N = 2, resulting in φ unreachable. µArchiFI: Formal Modeling and Verification Strategies for Microarchitectural Fault Injections
[3] The Tiny AES implementation by kokke provides the AES key schedule analyzed in the µARCHIFI case study. µArchiFI: Formal Modeling and Verification Strategies for Microarchitectural Fault Injections
[4] Recent work models the AES key schedule as a computational graph and uses a neural message-passing network to recover AES master keys from corrupted cold-boot memory, outperforming prior cold-boot attack methods. Recovering AES Keys with a Deep Cold Boot Attack