Back to Intelligence

SilentAir and LIMINAL PANDA: How Threat Actors Inside Mobile Networks Spy on Subscribers — Detection and Hardening Guide for Telecom Defenders

SA
Security Arsenal Team
November 16, 2026
10 min read

Telecom security firm Enea has published research on two distinct threat actors — SilentAir and LIMINAL PANDA — operating inside mobile operator networks to surveil subscribers, intercept voice calls, and capture SMS traffic. SilentAir is assessed to have compromised operators since before 2025, meaning we are looking at long-dwell-time intrusions in the most sensitive tier of national communications infrastructure.

This is not a theoretical concern. If an actor is positioned inside the mobile core — at the signaling layer — they don't need malware on a target's phone. They can redirect SMS (including MFA one-time passwords), intercept calls, track subscriber location, and profile call metadata at scale. For defenders at operators, MVNOs, and any enterprise that depends on SMS-based authentication, this campaign demands immediate attention. Organizations that run their own signaling infrastructure or peer with carrier networks should assume adversary interest in their interconnect points and hunt accordingly.

Technical Analysis

What We Know

  • SilentAir: Long-running intrusion campaign inside mobile operator networks, active since before 2025. Tradecraft is consistent with persistent, low-noise access to the mobile core — the goal is sustained surveillance capability, not disruption.
  • LIMINAL PANDA: A second, separate actor also operating within mobile networks, focused on subscriber espionage. The "PANDA" naming convention is widely used in the industry for China-nexus actors; Enea's research treats these as distinct groups with overlapping objectives.

How Mobile-Core Espionage Works (Defender's View)

Actors inside a carrier network typically weaponize the signaling plane rather than deploying commodity malware on endpoints. The classic playbook:

  1. Initial access and persistence on reachable core elements — often Linux-based nodes: signaling transfer points (STPs), Diameter routing agents (DRAs), HLR/HSS front-ends, SMSCs, or operational support systems (OSS/BSS) with weak segmentation and legacy credentials.
  2. Abuse of legitimate signaling protocols — SS7 (MAP/CAMEL) and Diameter — to issue subscriber queries. Techniques include fraudulent SendAuthenticationInfo, UpdateLocation, ProvideSubscriberInfo, and AnyTimeInterrogation requests to harvest location, or malicious call-forwarding/SMSC reconfiguration to duplicate SMS and voice to attacker-controlled endpoints.
  3. Eavesdropping enablement via lawful-intercept-style configurations or manipulation of CAMEL subscription data (O-CSI/T-CSI triggers), causing the network itself to fork media and messages to the adversary.
  4. Evasion through living-off-the-infrastructure: using native GGSN/SGSN tooling, valid OSS accounts, and encrypted management tunnels, producing almost no traditional endpoint telemetry.

Because much of the malicious behavior is protocol-legitimate (SS7/Diameter have no built-in authentication or encryption), detection depends on behavioral anomaly detection at the signaling firewall and rigorous host telemetry on the core nodes themselves.

Affected Platforms and Exposure

  • Mobile network operators (MNOs) and MVNOs with 2G/3G/4G cores (SS7/Diameter exposure). 5G standalone reduces — but does not eliminate — risk where legacy interworking functions remain.
  • Any enterprise relying on SMS for MFA or transaction verification: subscriber-level interception bypasses your endpoint controls entirely.
  • Exposure is highest where: interconnect partners are numerous, signaling firewalls are absent or misconfigured, and core nodes run unpatched Linux with permissive SSH access from vendor jump hosts.

Exploitation Status

Confirmed active, in-the-wild intrusions. SilentAir has persisted in operator environments since before 2025. No CVE is associated with this campaign — the tradecraft abuses architectural weaknesses and valid credentials rather than a single patchable bug. That makes detection and segmentation, not patch management alone, the primary defense.

Detection & Response

The detections below target the observable, host- and network-layer behaviors this tradecraft requires: unauthorized interactive access to core/signaling nodes, signaling-plane reconnaissance, and modification of subscriber or call-handling configuration. They are deliberately scoped to high-fidelity signals.

Sigma Rules

YAML
---
title: Interactive Shell Access to Telecom Core Signaling Node
id: 3f1a7c92-6b4d-4e18-9a2c-5d7e8f901234
status: experimental
description: Detects interactive shells spawned by SSH or management daemons on Linux telecom core nodes (STP, DRA, HSS, SMSC), consistent with hands-on access observed in mobile-core espionage campaigns such as SilentAir and LIMINAL PANDA.
references:
  - https://securityaffairs.com/203099/apt/silentair-and-liminal-panda-target-telecom-networks-to-spy-on-subscribers.html
  - https://attack.mitre.org/techniques/T1059/004/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.execution
  - attack.t1059.004
logsource:
  category: process_creation
  product: linux
detection:
  selection_parents:
    ParentImage|endswith:
      - '/sshd'
      - '/dropbear'
      - '/telnetd'
  selection_shells:
    Image|endswith:
      - '/bash'
      - '/sh'
      - '/dash'
      - '/zsh'
      - '/python'
      - '/python3'
      - '/perl'
  condition: selection_parents and selection_shells
falsepositives:
  - Legitimate vendor and NOC administrative sessions — baseline by source account and jump host, alert on deviations
level: high
---
title: Subscriber or Call-Handling Configuration Modification on Linux Core Node
id: 8c2d5e14-1a9b-4f37-b6d0-2e4a6c8d0f12
status: experimental
description: Detects tampering with configuration and subscriber-data files on telecom core elements, a technique used to enable SMS/call interception by altering routing, forwarding, or interception parameters.
references:
  - https://securityaffairs.com/203099/apt/silentair-and-liminal-panda-target-telecom-networks-to-spy-on-subscribers.html
  - https://attack.mitre.org/techniques/T1565/001/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.impact
  - attack.t1565.001
logsource:
  category: file_event
  product: linux
detection:
  selection_paths:
    TargetFilename|contains:
      - '/etc/ss7'
      - '/etc/diameter'
      - '/etc/sigtran'
      - 'smsc.conf'
      - 'hlr.cfg'
      - 'hss.cfg'
      - 'dra.conf'
      - 'camel'
  condition: selection_paths
falsepositives:
  - Planned network changes — correlate with change-management tickets and maintenance windows
level: high
---
title: Network Reconnaissance Tooling on Telecom Core Linux Hosts
id: 5b9e3f06-7d2a-4c81-a3e9-0f1b3d5c7a92
status: experimental
description: Detects packet capture and network scanning utilities executed on mobile-core nodes. Attackers surveilling signaling traffic (SS7/Diameter) commonly run tcpdump/tshark or scan for adjacent core elements.
references:
  - https://securityaffairs.com/203099/apt/silentair-and-liminal-panda-target-telecom-networks-to-spy-on-subscribers.html
  - https://attack.mitre.org/techniques/T1040/
author: Security Arsenal
date: 2026/04/06
tags:
  - attack.discovery
  - attack.t1040
  - attack.t1046
logsource:
  category: process_creation
  product: linux
detection:
  selection:
    Image|endswith:
      - '/tcpdump'
      - '/tshark'
      - '/wireshark'
      - '/nmap'
      - '/masscan'
      - '/ncat'
      - '/socat'
  condition: selection
falsepositives:
  - NOC troubleshooting captures — restrict to authorized accounts and alert on others
level: medium

Microsoft Sentinel / Defender KQL

This hunt assumes your telecom core nodes and signaling firewalls forward logs to Sentinel via Syslog/CEF — the standard pattern for carrier environments. It hunts for management-plane access to signaling nodes from unusual sources, plus authentication anomalies on core infrastructure.

KQL — Microsoft Sentinel / Defender
// Hunt: unusual administrative access and auth anomalies on telecom core/signaling nodes
// Ingestion: Syslog and/or CEF (CommonSecurityLog) from Linux core elements and signaling firewalls
let CoreHosts = dynamic(["stp", "dra", "hss", "hlr", "smsc", "mme", "sgw", "pgw", "ggsn", "sgsn"]);
let Window = 14d;
let KnownAdmins =
    SigninLogs // or your inventory of authorized NOC accounts/jump hosts — adjust to your CMDB
    | summarize by dummy = 1;
union isfuzzy=true
    (Syslog
    | where TimeGenerated > ago(Window)
    | where Computer has_any (CoreHosts)
    | where SyslogMessage has_any ("Accepted password", "Accepted publickey", "session opened")
    | extend SourceIP = extract(@"from ([0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3})", 1, SyslogMessage)
    | extend AuthUser = extract(@"for (?:invalid user )?(\S+) from", 1, SyslogMessage)
    | summarize Sessions=count(), FirstSeen=min(TimeGenerated), LastSeen=max(TimeGenerated)
        by Computer, SourceIP, AuthUser),
    (CommonSecurityLog
    | where TimeGenerated > ago(Window)
    | where DeviceProduct has_any ("firewall", "signaling") or DestinationHostName has_any (CoreHosts)
    | where DeviceAction in~ ("allow", "permit", "accept")
    | summarize Flows=count() by SourceIP, DestinationHostName, DestinationPort, DeviceVendor))
| order by LastSeen desc

Pair this with a second-pass query that flags first-seen source IPs touching signaling ports (SCTP 2905 for M3UA/SIGTRAN, 3868 for Diameter, 2904/2906) — in a well-baselined core, new sources on those ports are nearly always interesting.

Velociraptor VQL

For IR scoping on a suspect Linux core node, this artifact pulls listening processes, established SCTP connections (signaling traffic rides SCTP, not TCP/UDP — a fact many EDR stacks ignore), and persistence artifacts in one sweep.

VQL — Velociraptor
-- Telecom Core Node Triage: signaling connections, listeners, and persistence
-- Targets Linux core elements (STP/DRA/SMSC/HSS) suspected of SilentAir/LIMINAL PANDA-style compromise

-- 1) Processes with active network connections (catches SCTP listeners via /proc)
SELECT Pid, Name, CommandLine, Username, Exe
FROM pslist()
WHERE Exe =~ '(sshd|dropbear|socat|ncat|tcpdump|tshark|python|perl)'
   OR CommandLine =~ '(2899|2904|2905|3868|sctp|diameter|ss7|sigtran)'

-- 2) Raw SCTP socket state — signaling plane visibility most tools miss
SELECT laddr, lport, raddr, rport, status, inode
FROM parse_file(filename='/proc/net/sctp/assocs',
                accessor='proc',
                separator=' ')

-- 3) Persistence sweep: cron, systemd units, and authorized_keys modified in last 90 days
SELECT FullPath, Mtime, Size
FROM glob(globs=['/etc/cron.d/*', '/etc/crontab', '/var/spool/cron/*',
                 '/etc/systemd/system/*.service', '/root/.ssh/authorized_keys',
                 '/home/*/.ssh/authorized_keys'])
WHERE Mtime > (now() - 90 * 24 * 3600)
ORDER BY Mtime DESC

Hardening & Verification Script

This Bash audit script is for Linux-based telecom core nodes. It verifies exposure of signaling ports, audits recent auth anomalies, checks for sniffers, and validates authorized_keys integrity — the highest-value checks for this threat class.

Bash / Shell
#!/bin/bash
# Telecom Core Node Audit — SilentAir / LIMINAL PANDA class intrusions
# Run as root on STP/DRA/SMSC/HSS/MME nodes. Read-only; produces an audit report.
set -u
REPORT="/var/log/core-audit-$(hostname)-$(date +%Y%m%d).log"
exec > >(tee -a "$REPORT") 2>&1

echo "=== [1] Signaling port exposure (should only listen on expected interfaces) ==="
ss -lntup | grep -E ':(2899|2904|2905|2906|3868|5060|22)\b' || echo "No unexpected listeners"

echo "=== [2] SCTP associations (SS7/Diameter ride SCTP) ==="
cat /proc/net/sctp/assocs 2>/dev/null || echo "No SCTP module loaded"

echo "=== [3] Recent successful logins — review for non-NOC sources ==="
last -i | head -30
grep -E "Accepted (password|publickey)" /var/log/secure* /var/log/auth.log* 2>/dev/null | tail -40

echo "=== [4] Sniffers and recon tooling present ==="
which tcpdump tshark nmap masscan socat ncat 2>/dev/null
pgrep -al 'tcpdump|tshark|nmap|socat|ncat' || echo "None running"

echo "=== [5] SSH key persistence review ==="
for k in /root/.ssh/authorized_keys /home/*/.ssh/authorized_keys; do
  [ -f "$k" ] && echo "--- $k (mtime: $(stat -c %y "$k")) ---" && cat "$k"
done

echo "=== [6] Recently modified core configs (last 30 days) ==="
find /etc -name '*ss7*' -o -name '*diameter*' -o -name '*sigtran*' -o -name 'smsc*' -o -name 'dra*' 2>/dev/null | xargs -r find -mtime -30 -ls

echo "=== [7] Unauthorized world-writable or new SUID binaries ==="
find /usr /opt -perm -4000 -mtime -90 2>/dev/null | head -20

echo "=== Audit complete: $REPORT — ship to SIEM and diff against baseline ==="

Remediation

There is no patch for this campaign — remediation is architectural and operational. Prioritize in this order:

  1. Deploy and enforce a signaling firewall / SS7-Diameter security gateway (Enea, Cellusys, Positive Technologies, or equivalent) at all interconnect boundaries. Block Category 2/2.5/3 messages per GSMA FS.11 (SS7) and FS.19 (Diameter) — specifically ProvideSubscriberInfo, AnyTimeInterrogation, and UpdateLocation originating from foreign networks where the subscriber is not roaming there.
  2. Audit interconnect partners and roaming links. SilentAir-class actors frequently enter via a weaker peer. Inventory every SS7/Diameter association, terminate dormant ones, and require mutual security attestations.
  3. Segment the management plane. Core nodes must only accept SSH from hardened jump hosts with MFA and session recording. Rotate all credentials and SSH keys on core elements now — assume pre-2025 compromise of any credential older than that.
  4. Establish a signaling baseline, then alert on deviation. New SCTP associations, first-seen GTs (Global Titles), and out-of-hours configuration changes on STP/DRA/SMSC elements should page a human.
  5. File integrity monitoring on core configs (SS7/Diameter routing tables, SMSC forwarding rules, CAMEL subscription data). Any unplanned change is an incident until proven otherwise.
  6. For enterprises: stop using SMS for MFA and high-value transaction verification. Move to FIDO2/passkeys or authenticator-based TOTP. Assume SMS delivery to high-value targets is interceptable at the carrier layer — because, per Enea's findings, it is.
  7. Engage national CERTs and GSMA's CVD program if you find evidence of compromise; telecom intrusions of this type are typically multi-operator and require coordinated response.

The strategic lesson: adversaries have concluded — correctly — that the mobile core is softer than the endpoint. If your threat model stops at the enterprise perimeter, you have a blind spot the size of the public telephone network.

Related Resources

Security Arsenal Incident Response Services AlertMonitor Platform Book a SOC Assessment incident-response Intel Hub

Is your security operations ready?

Get a free SOC assessment or see how AlertMonitor cuts through alert noise with automated triage.