ARCEP 2026 Observatory: 43% Phone Spoofing, BBR Replaces Cubic, and What It Means for Field Diagnostics

ARCEP’s 2026 observatory on telecommunications services quality in France delivers two findings that deserve close attention from network professionals: 43% of French consumers have been affected by phone number spoofing, and ARCEP is shifting its official speed measurement methodology from Cubic to BBR congestion control. Both developments have direct implications for field diagnostics, network measurement accuracy, and the security architecture of mobile voice services.

This article provides a technical analysis of both findings, their interconnection through the signaling layer, and what RF engineers and field teams need to understand about the changing measurement landscape.

---

The 2026 observatory: key findings

Consumer satisfaction metrics

Metric	Value	Year-over-Year Change
Overall telecom satisfaction	7.9/10	+0.3
Mobile satisfaction	7.8/10	+0.2
Fixed broadband satisfaction	8.0/10	+0.4
Orange satisfaction (highest)	8.1/10	Stable
Free Mobile satisfaction	7.6/10	+0.3
SFR satisfaction	7.5/10	+0.2
Bouygues Telecom satisfaction	7.7/10	+0.2
Improvement over 2 years	+7 points (percentage)	Significant trend

The 7.9/10 average satisfaction and the 7-point improvement over two years indicate that French operators have materially improved their service quality. Orange’s leadership at 8.1/10 reflects its infrastructure investment advantage (largest fiber and 4G/5G footprint in France).

Phone spoofing: the 43% statistic

The headline finding: 43% of French consumers report having been affected by phone number spoofing (usurpation de numero / spoofing d’identite d’appel). This includes:

• Receiving calls displaying a trusted number (bank, government agency, utility) that is actually a scam
• Having their own number spoofed by fraudsters calling third parties
• Receiving SMS from spoofed sender IDs

This 43% figure represents a massive security failure in the telecommunications signaling infrastructure, not in the IP layer or the radio layer, but in the SS7 and SIP signaling layer that routes and identifies voice calls.

---

Phone spoofing: the signaling vulnerability

How caller ID spoofing works

Phone number spoofing exploits the trust model of the Public Switched Telephone Network (PSTN) and its VoIP successor architecture. The fundamental vulnerability exists at two levels:

SS7-level spoofing (legacy interconnect)

Calling Party                SS7 Network              Called Party
    │                            │                          │
    ├── IAM (Initial Address  ──►│                          │
    │   Message) with forged     │                          │
    │   Calling Party Number     │──── IAM forwarded ──────►│
    │                            │     with forged CPN      │
    │                            │                          │
    │                            │     No verification      │
    │                            │     of CPN authenticity   │

The SS7 Initial Address Message (IAM) carries the Calling Party Number (CPN) field. In the original SS7 architecture (designed in the 1970s-1980s), there is no authentication of the CPN. Any SS7 interconnection point can inject an IAM with any CPN value. This design assumed that all SS7 network participants were trusted telephone companies. That assumption has not held since the deregulation of telecom in the 1990s.

SIP-level spoofing (VoIP/IMS interconnect)

SIP INVITE sip:called@operator.fr SIP/2.0
From: <sip:+33144001234@spoofed-origin.com>;tag=xyz
P-Asserted-Identity: <sip:+33144001234@trusted-carrier.fr>

In SIP-based voice (including VoLTE), the From header and P-Asserted-Identity header carry the calling number. While SIP does support identity assertion mechanisms (P-Asserted-Identity within trusted networks), the trust boundary between operators is frequently insufficient. A rogue SIP trunk provider can inject calls with arbitrary identity headers.

The STIR/SHAKEN response

The industry response to caller ID spoofing is STIR/SHAKEN (Secure Telephony Identity Revisited / Signature-based Handling of Asserted information using toKENs):

Component	Function	Status in France (2026)
STIR	Cryptographic framework for signing caller ID	Specification complete (RFC 8224, 8225, 8226)
SHAKEN	Governance and certificate authority framework	US: mandatory since 2021. France: voluntary, adoption in progress
Attestation Level A	Full attestation: carrier confirms customer identity	Required for direct enterprise connections
Attestation Level B	Partial attestation: carrier authenticated call origin, not specific caller	Most inter-carrier calls
Attestation Level C	Gateway attestation: carrier received call from gateway, cannot verify origin	International/legacy interconnect

ARCEP’s 2026 observatory implicitly confirms that STIR/SHAKEN adoption in France remains incomplete. The 43% spoofing impact rate is incompatible with full Attestation A deployment. The French implementation timeline has been slower than the US mandate, largely because:

1. The European regulatory framework (EECC) does not mandate STIR/SHAKEN specifically
2. French operators interconnect with legacy SS7 networks across Europe, Africa, and the Middle East where no STIR/SHAKEN exists
3. The certificate authority governance for European STIR/SHAKEN is still being harmonized

What this means for field diagnostics

Phone spoofing is not a radio problem, but it manifests through the radio network. When field engineers investigate voice quality complaints, the diagnostic workflow must now include signaling verification:

New diagnostic questions for VoLTE:

• Is the P-Asserted-Identity consistent with the SIP From header?
• Does the SIP INVITE carry STIR/SHAKEN attestation headers?
• What attestation level was asserted, and by which certificate authority?
• Are there anomalous SIP headers indicating transit through untrusted intermediaries?

VoLTE diagnostic tools that capture SIP signaling (INVITE, 180 Ringing, 200 OK, BYE) already have the raw data needed to identify spoofing indicators. The analysis layer needs to evolve to flag identity inconsistencies as part of the standard call quality diagnostic.

---

BBR replaces Cubic: the measurement revolution

Background: what changes and why

ARCEP’s speed measurement methodology has historically used TCP Cubic as the default congestion control algorithm. Cubic, developed at NCSU and adopted as Linux’s default in 2006, has been the de facto standard for internet speed measurement for nearly two decades.

In the 2026 methodology update, ARCEP signals a shift toward BBR (Bottleneck Bandwidth and Round-trip propagation time), developed by Google and increasingly deployed on major content delivery networks.

Cubic vs. BBR: technical differences

Characteristic	Cubic	BBR
Congestion signal	Packet loss	Bandwidth estimation + RTT measurement
Behavior on loss	Multiplicative decrease (window reduction)	Models bottleneck bandwidth; less reactive to loss
Performance on lossy links	Degrades significantly (e.g., -30% at 1% loss)	Maintains throughput despite moderate loss
Performance with bufferbloat	Fills buffers, creates latency	Actively manages inflight data to minimize latency
Fairness with Cubic flows	Fair (same algorithm)	BBR can dominate Cubic flows on shared links
Mobile network performance	Penalized by radio link retransmissions	Better adapts to variable radio conditions
Default on Google servers	No (migrated to BBR)	Yes (YouTube, Google Cloud, etc.)
Default on Linux kernel	Yes (since 2006)	Available since 4.9 (2016), increasingly adopted

The measurement gap: 10-30%

The switch from Cubic to BBR is not a cosmetic change. Field measurements demonstrate a 10-30% throughput difference between the two algorithms on the same network path at the same time:

Network Condition	Cubic Throughput	BBR Throughput	Difference
Clean path (0% loss, low RTT)	100 Mbps	102 Mbps	+2% (negligible)
Moderate loss (0.5%)	85 Mbps	98 Mbps	+15%
High loss (1-2%)	55 Mbps	88 Mbps	+60%
Bufferbloated path	70 Mbps (high latency)	95 Mbps (lower latency)	+36%
LTE with handover	60 Mbps (drops during HO)	78 Mbps (recovers faster)	+30%
5G mmWave (variable)	800 Mbps (volatile)	950 Mbps (smoother)	+19%

These differences are not theoretical. They are measured. And they have profound implications for field diagnostics:

1. Historical comparison is broken. An operator’s 2025 speed measurements (Cubic-based) cannot be directly compared with 2026 measurements (BBR-based). A 15% throughput “improvement” may be entirely algorithmic, not network-related.

2. Field tools must support algorithm selection. A diagnostic tool that only runs Cubic speed tests in a BBR-measured regulatory environment produces non-comparable results. Professional tools need configurable congestion control.

3. VoLTE/video QoE correlation changes. If BBR shows higher throughput than Cubic on the same link, the correlation between speed test results and application QoE may shift. An RSRP/SINR condition that previously correlated with “insufficient throughput for HD video” under Cubic may show sufficient throughput under BBR.

4. Regulatory compliance requires algorithm awareness. When ARCEP publishes coverage statistics using BBR-measured throughput, operators must ensure their own internal measurements use the same algorithm to maintain comparability.

Why ARCEP is making the switch

ARCEP’s motivation is straightforward: BBR better represents actual user experience on modern networks.

The reasoning:

• 70%+ of internet traffic originates from servers running BBR (Google, YouTube, Netflix CDN, major cloud providers)
• Measuring with Cubic while users experience BBR creates a systematic negative bias in official statistics
• BBR’s resilience to packet loss better reflects the real-world performance of mobile networks where radio retransmissions are frequent
• The shift aligns France with international trends (FCC, Ofcom are evaluating similar methodology updates)

---

The intersection: signaling security and measurement integrity

Phone spoofing and BBR adoption may seem unrelated, but they share a common underlying theme: the signaling and protocol layers that field diagnostics must capture are becoming more complex, not less.

The expanding diagnostic scope

Year	Field Diagnostic Focus	Protocol Layers
2015	RF coverage + throughput	L1 (RF) + L7 (speed test)
2020	RF + signaling + QoE	L1 + L3 (RRC/NAS) + L7 (VoLTE MOS)
2026	RF + signaling + QoE + security + algorithm-aware measurement	L1 + L3 + L7 + SIP security headers + TCP CC selection

Field engineers in 2026 need tools that:

1. Measure RF parameters at engineering precision (RSRP, RSRQ, SINR via chipset interface)
2. Decode Layer 3 signaling in real time (RRC, NAS)
3. Capture VoLTE SIP signaling for quality and security analysis
4. Execute speed tests with configurable congestion control (Cubic and BBR)
5. Correlate all layers with GPS positioning for spatial analysis
6. Export in formats accepted by regulators and vendors

This is not a wish list. It is the minimum capability set for professional field diagnostics in the ARCEP 2026 framework.

---

Orange at 8.1/10: what the leader does differently

Orange’s leading satisfaction score (8.1/10) is not accidental. It reflects structural advantages that field diagnostics help quantify:

Factor	Orange Advantage	Measurable With Field Tools
Fiber coverage	35M+ premises connectable	N/A (fixed)
4G population coverage	99.4%	RSRP/RSRQ mapping, coverage validation
5G NR deployment	8,000+ sites (SA + NSA)	NR-RSRP, SSB beam measurement, EN-DC monitoring
VoLTE quality	Highest MOS scores in benchmarks	VoLTE MOS per call, codec analysis
Handover success rate	>98.5% in drive test benchmarks	Layer 3 handover event analysis
Indoor coverage	DAS in major venues, small cells in enterprises	Indoor walk test, serving cell identification

The operators trailing Orange (SFR at 7.5, Free at 7.6) have specific, measurable gaps that field diagnostics can identify and quantify. The satisfaction gap is not abstract; it maps to RF parameters, signaling events, and QoE metrics that professional tools capture.

---

Practical implications for field teams

Updated measurement protocol for 2026

Given ARCEP’s methodology changes, field teams should update their measurement protocols:

Speed testing:

1. Execute speed tests using both Cubic and BBR congestion control
2. Record the algorithm used with every measurement
3. Use BBR results for ARCEP regulatory comparison
4. Use Cubic results for historical trend analysis (maintaining continuity with pre-2026 data)
5. Document the throughput delta between algorithms at each measurement point

VoLTE diagnostics:

1. Capture full SIP signaling (INVITE through BYE) for every test call
2. Check P-Asserted-Identity and STIR/SHAKEN attestation headers
3. Flag calls with Attestation Level C or missing attestation
4. Record MOS, jitter, and codec for quality baseline
5. Document any identity header inconsistencies for security audit

RF measurement:

1. No change to RSRP/RSRQ/SINR methodology
2. Add NR-specific measurements for 5G sites (SS-RSRP, SS-RSRQ, SS-SINR)
3. Record SSB beam index for 5G beam management analysis
4. Capture CA activation status and component carrier configuration

Export requirements for ARCEP compliance

Data Point	Format Requirement	Sampling Rate
GPS position	WGS84, sub-10m accuracy	Per measurement point
RSRP/RSRQ/SINR	Per cell (serving + neighbors)	Every 100ms minimum
Throughput	Per-second with algorithm noted	Continuous during test
VoLTE MOS	Per call	Every call
Timestamp	UTC with millisecond precision	Every data point
Cell ID (PCI, EARFCN, NR-ARFCN)	Per serving and neighbor	Every 100ms minimum

---

The broader European context

ARCEP’s methodology update is not isolated. European regulators are converging on modernized measurement approaches:

Regulator	Country	BBR Adoption Status	Spoofing Mitigation
ARCEP	France	Adopted (2026)	STIR/SHAKEN voluntary adoption
Ofcom	UK	Under evaluation	CLI authentication mandate (2025)
BNetzA	Germany	Pilot phase	National CLI verification program
AGCOM	Italy	Not yet announced	EU-level coordination
CNMC	Spain	Not yet announced	EU-level coordination
BIPT	Belgium	Pilot phase	National spoofing task force

The trend is clear: European regulators are simultaneously modernizing speed measurement methodology and addressing signaling security. Field diagnostic tools that support both dimensions will be better positioned for regulatory compliance across multiple markets.

---

Conclusion

ARCEP’s 2026 observatory delivers two messages that the network measurement community cannot ignore. The 43% phone spoofing rate exposes a signaling security failure that extends into VoLTE diagnostic requirements. The Cubic-to-BBR transition creates a 10-30% measurement discontinuity that every field team must account for.

Both developments point in the same direction: field diagnostics in 2026 requires more protocol depth, more configurability, and more security awareness than ever before. The RF layer alone is no longer sufficient. Layer 3 signaling, SIP header analysis, and congestion control algorithm selection are now part of the professional diagnostic toolkit.

Orange’s 8.1/10 satisfaction leadership and the 7-point improvement across all operators over two years demonstrate that French mobile networks are improving. The tools used to measure and validate that improvement must evolve at the same pace.

For field engineers and network planners, the ARCEP 2026 observatory is not just a regulatory update. It is a recalibration of what “measurement” means: deeper into the signaling stack, more aware of security, and more precise about the algorithms that determine reported performance.