Hunting c2 beaconing with frequency analysis

Hunting C2 Beaconing with Frequency Analysis

When to Use

• You have connection logs (proxy, firewall, Zeek/conn.log, NetFlow) and want to find implants that call home on a schedule.
• You are testing a hypothesis that a beacon is hiding in normal-looking web traffic.
• You need to rank source→destination pairs by timing regularity, allowing for jitter.

Do not use pure interval regularity as a verdict — software update checks, telemetry, and keep-alives also beacon; combine timing with destination reputation and data volume.

Prerequisites

• Connection records with timestamp, source, destination (IP/domain), and ideally bytes, over a window long enough to capture many callbacks (hours to days).
• A way to enrich destinations (reputation, age, rarity) for follow-up.

Workflow

Step 1: Group and order by pair

Bucket connections by (source, destination) and sort timestamps. Require a minimum count so the interval statistics are meaningful.

Step 2: Compute inter-arrival intervals

Derive deltas between consecutive connections per pair; the interval distribution reveals periodicity.

Step 3: Score regularity with jitter tolerance

A low coefficient of variation (std/mean) of intervals indicates a steady beacon; modern beacons add jitter, so score on tolerance rather than requiring identical intervals.

python scripts/analyst.py beacon conn.json --min-events 8

Step 4: Reduce false positives

Down-rank known update/telemetry destinations and CDNs; up-rank rare/young domains, small fixed payload sizes, and odd ports.

Step 5: Triage and confirm

For top pairs, pull payloads/JA3, destination intel, and host context; confirm via the C2/beacon config skills and escalate.

Validation

• Top candidates show consistently spaced callbacks (low CV) over many events, not a handful.
• Known-benign periodic services are filtered or explained.
• Confirmed beacons corroborate with destination reputation or payload analysis.

Pitfalls

• Too few events per pair, making the interval statistics noise.
• Requiring perfect periodicity and missing jittered beacons.
• Ignoring data-volume regularity, a strong secondary beacon signal.

References

• See references/api-reference.md for the beacon scorer.
• ATT&CK T1071 and inter-arrival timing analysis (linked in frontmatter).