3GPP TR 33.809

Study on 5G Security Enhancement against False Base Stations (FBS)
V18.0.0 (2023-12) — Release 18 Technical Report

1. Scope

▶

Studies threats from false base stations (IMSI catchers, Stingrays, rogue gNBs) against 5G NR, analyzes existing 5G protections, identifies remaining gaps, and proposes enhancements.

Real-world impact: False base stations (FBS) are actively used by law enforcement, intelligence agencies, and criminals worldwide. While 5G introduced SUCI encryption and improved integrity protection, this study finds residual vulnerabilities remain.

2. False Base Station Attack Taxonomy

▶

2.1 IMSI/SUPI Catching

▶

Attack: Identity Capture

Legacy (2G-4G): FBS sends Identity Request for IMSI → UE responds with plaintext IMSI.

5G mitigation: SUCI (Subscription Concealed Identifier) encrypts SUPI using home network public key (ECIES). FBS cannot decrypt.

Residual risk: 5G-GUTI reallocation patterns may allow correlation; SUCI scheme 0 (null encryption) still allowed in some configs.

2.2 Denial of Service

▶

Attack: RRC Reject / Redirect

FBS broadcasts stronger signal than legitimate gNB → UE camps on FBS → FBS sends RRC Reject or redirect to another frequency, denying service.

5G status: RRC Reject before security activation is not integrity protected. UE cannot verify legitimacy.

Attack: Registration Reject with Specific Causes

FBS impersonates AMF response via NAS Registration Reject with cause #3 (Illegal UE), #6 (Illegal ME), or #7 (5GS services not allowed).

Impact: UE enters deregistered state; may require manual intervention or USIM removal to recover.

5G status: TS 33.501 §6.1.4 — NAS Reject before authentication is not integrity protected. Timer-based recovery (T3247) added as partial mitigation.

2.3 Bidding-Down / Downgrade Attacks

▶

Attack: Force Fallback to 4G/3G/2G

FBS broadcasts that 5G is unavailable → UE falls back to 4G/3G → loses 5G security features (SUCI, UP integrity).

Residual risk: While Anti-bidding-down between architectures (ABBA) parameter exists in 5G, UE cannot verify "no 5G coverage" claims before falling back.

2.4 MITM Relay

▶

Attack: Full MITM Relay (Most Sophisticated)

FBS relays all messages between UE and legitimate gNB in real-time, acting as transparent proxy.

Capability: Can selectively modify unprotected messages, inject SIB changes, perform measurement manipulation.

5G countermeasure: Integrity protection of NAS and RRC prevents modification of protected messages, but pre-security messages remain vulnerable.

3. Existing 5G Protections

▶
ProtectionAddressesSpec Reference
SUCI EncryptionIMSI catchingTS 33.501 §6.12
NAS Integrity (mandatory)NAS message modificationTS 33.501 §6.7
RRC Integrity (mandatory)RRC message modificationTS 33.501 §6.8
UP Integrity (optional)User data modificationTS 33.501 §6.6
ABBA ParameterBidding-down between architecturesTS 33.501 §6.1.3
T3247 TimerNAS Reject recoveryTS 24.501

4. Remaining Gaps Identified

▶
GapExploitable BySeverity
SIB (System Info Broadcast) not integrity protectedFBS broadcasting fake cellsHigh
RRC Reject before security — no integrityFBS DoS on connection setupHigh
Paging not authenticatedFBS paging to lure UEsMedium
Cell reselection based on signal strengthFBS with stronger signal winsHigh
Measurement reports before AS securityMeasurement manipulation for trackingMedium
SUCI scheme 0 (null) still permittedIMSI catching when scheme 0 configuredMedium

5. Proposed Solutions

▶
SolutionApproachStatus
SIB Integrity ProtectionDigital signature on SIB1/SIB2 using network public keyUnder study — compute cost concern
Multi-SIM FBS DetectionUE uses multiple SIMs/radios to cross-check cell infoProposed — device cost increase
Crowdsourced DetectionUEs report suspicious cells to network for analysisRel-18 contributions
Cell AuthenticationgNB proves identity to UE before initial accessFundamental redesign — long-term
AI-based DetectionUE/network ML models detect FBS patternsResearch phase
Key conclusion: Complete FBS protection requires authenticating the cell to the UE before any signaling — fundamentally inverting today's model where only the UE authenticates to the network. This remains an open research challenge.