ACSE — The Architecture of Continuous Cryptographic Motion

Abstract

Almost every significant cyber breach of the past decade shares a structural precondition: the attacked system's observable surface remained static and predictable long enough for the attacker to build a usable map. Firewalls, encryption, and zero-trust architectures protect known surfaces but cannot defend surfaces that remain unchanged throughout an attacker's reconnaissance cycle.

Adaptive Cryptographic Surface Engineering (ACSE) is designed to remove this precondition for the surfaces it covers. The Kali Invariant — ACSE's central formal property — ensures that every covered surface's fingerprint changes at every rotation cycle (60 s by default). The fingerprint is computed from fresh entropy; what an attacker can observe of a real host changes at every rotation and, for the host TCP/IP stack, cannot be predicted better than chance within its 48 states (measured).

1. The Reconnaissance Advantage

1.1 The Attacker's Primary Asset

An attacker's ability to exploit a target depends not on a single moment of observation but on the accumulation of stable observations over time. API endpoints, database connection identifiers, session tokens, network topology fingerprints, and TLS certificate identifiers all remain valid — in most systems — for durations measured in hours, days, or months.

This stability is the attacker's primary asset. A credential stolen today is usable tomorrow. A network map built over three weeks remains accurate next month. The attacker's investment in reconnaissance compounds: each observation enriches a model that was not available to previous observations.

1.2 Why Existing Defences Do Not Solve This

• Static defences (firewall, IDS, encryption): Protect the contents of surfaces but leave the surfaces themselves observable and stable.
• Moving Target Defence (MTD): Rotates surfaces on a schedule — typically hourly or daily. The rotation interval is the vulnerability: an attacker observing within any window builds a stable model. No cryptographic proof of independence between successive states is provided.
• Zero Trust Architecture: Verifies identity continuously but treats the identity itself as static. A compromised credential gives the adversary a static, verifiable identity that zero-trust systems authenticate correctly.

2. The Kali Invariant

2.1 Formal Statement

Surface S: Any observable security boundary — an API endpoint, session token, network fingerprint, or cryptographic identity — that an attacker can probe or model.

Observable fingerprint F(S,t): The complete set of observable properties of surface S at time t.

Reconnaissance advantage A(t₁,t₂): The conditional probability that an observation at t₁ provides exploitable information about surface state at t₂, where t₂ > t₁.

This bounds the reconnaissance advantage; it does not remove a vulnerability that an exploit targets directly. A flaw reachable at a stable endpoint — an unpatched service, an injection bug — remains exploitable regardless of fingerprint rotation, so ACSE-PME is a layer for the reconnaissance phase, not a substitute for patching or input validation. The Kali Invariant is not probabilistic: it is a structural property of the fingerprint, enforced at every rotation cycle and, on the surfaces where a verifier runs, checked from outside by an independent verifier. The security rests on entropy quality, not the secrecy of the mechanism.

3. ACSE Architectural Model

3.1 Three-Layer Stack

3.2 KaliCore — Governing Intelligence

KaliCore is the governing intelligence of the mutation layer. It comprises three co-operating subsystems: EntropyManager (sensory — supplies entropy), EWMA Anomaly Scorer + State Machine (cognitive — decides mutation aggression), and Scheduler + StateTracker (executive — orchestrates timing and rollback).

KaliCore operates in three escalating states: Sachs (steady-state, EWMA ≥ 0.00–0.35), Hunter (elevated, ≥ 0.35), and Main Organ (full response, ≥ 0.70). Transitions are autonomous — no human intervention required.

4. The Atomic Mutation Cycle

Every mutation, regardless of profile, follows exactly four phases:

Phase 3 is the invariant enforcement point. If validation fails, the cycle rolls back to the Phase 1 checkpoint atomically. The Kali Invariant is never violated by a failed mutation attempt.

5. Organ State Machine

6. Security Properties

6.1 Fingerprint Independence

By the Kali Invariant and SHA3-256 preimage resistance, successive surface fingerprints are computationally independent. Computing F(S,t+1) without the entropy input requires inverting SHA3-256.

6.2 Past States After a Compromise

Each cycle's state is derived from fresh entropy rather than from the previous state. Whether a compromise of the engine's entropy generator exposes earlier states has not been assessed, so no forward-secrecy property is claimed.

6.3 Audit Chain Integrity

The SHA3-256 chained audit log provides tamper evidence for the complete mutation history. Any modification changes the chain hash of that record, invalidating all subsequent records.

6.4 Peer Authentication Secrecy

In ProVerif, an adversary who controls the network cannot learn the ally channel fingerprint, and each completed exchange authenticates both peers. This is a secrecy and authentication result, not a zero-knowledge proof. See WP-07 for the analysis results.

7. Comparison with Prior Approaches

8. Conclusion

ACSE addresses the fundamental precondition of modern cyber attacks: the stability of observable surfaces. The Kali Invariant is a structural property enforced at every rotation cycle and checked by an independent verifier. The attacker's map of the host goes out of date at the next rotation.