Build plan · revised 2026-11-10

Home network audit plan

A repeatable way to learn what every device on a home and lab network collects, what it sends out, and whether the two are linked. Each audit ends as a firewall allowlist.

What the source video showed

The starting point is The Homeboy Workshop's tap of a rooted LG OLED. The TV sat on an isolated segment behind an inline hardware tap. A laptop answered DNS and HTTP as a fake internet, with its own certificate authority, and advertised fake devices. The numbers below come from the video's transcript and description.

Most TLS handshakes never completed

Handshakes by outcome, fake internet with a private CA

About 9 in 10 conversations stopped at the certificate. The capture mostly records a TV failing to phone home, so it cannot show what the TV sends when it succeeds.

The TV questioned fake devices directly

Unicast requests to canaries, "in a few hours"

Canaries caught what a tap alone would only show as packets. This is the most reusable idea in the video.

Discovery repeats on a timer, and the stated mDNS rate does not match its own count

One hour of events, drawn from the rates the video gives

The video says mDNS queries came "roughly every 90 seconds", but it counts 57 in 4 hours, which is one per 253 s. This plan records every rate as a count and a window.

Gaps, and what this plan changes

Gap in the videoChange here
One network condition. The fake internet stops most conversations at the certificate.Three conditions: real internet, fake internet, no answer.
Collection is shown. Exfiltration is not.Unique canary tokens, plus repeated A/B upload comparison.
Treats the server's private key as the route to plaintext. Most TLS 1.2 is also ephemeral, and the vendor's key is never available.Plaintext at the endpoint: device key log, a proxy the device accepts, or metadata only.
One capture, one device state.A scripted state matrix.
No radio visibility, so the Wi-Fi scan claim stayed open. No mention of IPv6.Radio sniffers for management frames, BLE and 802.15.4. IPv6 captured and attributed by MAC.
One inline tap per device.A whole-home sensor first. The tap rig becomes the second tier.

Where to look

No single capture point answers all three questions. A capture on the router trunk sees everything that is routed, but not two devices talking inside one VLAN on one switch. The plan combines five sources.

Capture points on a segmented home network

trunk, all VLANs mirror or tap port 53 redirected here one per VLAN not on the trunk Internet Router / firewallNAT, VLAN gateway Resolverper-client query log Core switchtrunk and access ports Sensorpackets and Zeek logs Canariesfake devices, tokens Access pointIoT SSID, isolated IoT devicewired, VLAN iot IoT devicewired, VLAN iot Bench deviceVLAN bench Fake internetDNS, HTTP, bench CA 1 2 3 4 5 capture point, numbered as in the list below unicast inside one VLAN on one switch tier 2 bench only
  1. 1Trunk mirror or tap. Everything routed, before NAT, plus broadcast and link-local multicast from every VLAN.
  2. 2Resolver logs. Every name each device asks for, including reverse-lookup sweeps. No tap needed.
  3. 3Canaries. Unicast probes inside a VLAN, which the trunk never carries.
  4. 4Access point uplink. Wi-Fi clients' traffic as plain Ethernet. Take Wi-Fi payload here, not from the air.
  5. 5Bench tap or port mirror. One device, both directions, every packet.
Two more sources are not drawn: radio sniffers for Wi-Fi management frames, BLE, Zigbee and Thread, and host agents on machines you control.

What each source can see

BehaviourTrunk captureResolver logsCanariesPort mirror or bench tapRadio snifferHost agent
sees it sees part of it blind
Partial cells: resolver logs give names without volumes. Canaries and host agents see only probes aimed at them. A bench tap gives decrypted payload only with device keys or a proxy the device accepts. A host agent applies only to a machine you control.

What a passive capture can read when TLS is on

LayerFieldReadableLimit
2MAC addresses, VLAN tagYesIdentifies the device on its own segment.
3IP addressesYesIPv6 privacy addresses rotate. Attribute by MAC.
4Ports, packet sizes, timingYesEnough for volumes and beacon periods.
7DNS namesYesNot when the device uses DNS over HTTPS or TLS.
TLSServer name, client fingerprintYesEncrypted Client Hello hides the server name.
TLSServer certificateTLS 1.2 onlyTLS 1.3 encrypts it.
7PayloadNoNeeds keys from the device, or a proxy the device trusts.

Tier 1  always on

Whole-home sensor

Tier 1 watches every device cheaply and all the time. It answers question 2 in metadata, catches discovery and probing, and tells you which devices deserve the bench.

Phase 0

Inventory

  • Export DHCP leases and the router's ARP and NDP tables.
  • Build one table: device, owner, MAC, IPv4, IPv6 addresses, VLAN, switch port or SSID, radio protocols, rooted or not, firmware version.
  • Include the network gear itself: router, switches, access points, controller.
  • Give every audited device a DHCP reservation. Turn off MAC randomization on test devices only.

Done whenEvery MAC seen in a 7-day lease log maps to a row.

Aside: MikroTik RouterOS
  • Leases: /ip dhcp-server lease. Neighbours: /ip arp, /ipv6 neighbor.
  • MAC to switch port: /interface bridge host.
Aside: Ubiquiti UniFi, for comparison
  • The Client Devices list in UniFi Network already joins MAC, address, VLAN, and switch port or access point and SSID. On RouterOS you join three tables yourself.
  • Set a fixed IP per client in that list. Pull the list through the controller API if the interface offers no export.

Phase 1

Segmentation and DNS

  • VLANs: trusted, iot, lab, guest, mgmt, bench.
  • Default deny between VLANs. Log denied attempts.
  • Repeat every rule for IPv6. It has its own firewall tables on most platforms.
  • Run a local resolver with per-client query logs.
  • Redirect outbound port 53 to the resolver. Block outbound 853. Log connections to known DNS-over-HTTPS endpoints.
  • Disable UPnP, or leave it on for one week first and record who asks for port mappings.

Done whenResolver logs show per-device queries, and a device with a hard-coded resolver shows up in the redirect counters.

Aside: MikroTik RouterOS
  • VLANs with bridge VLAN filtering.
  • DNS redirect: dst-nat rules in /ip firewall nat for UDP and TCP 53, with the resolver's own address excluded.
  • The built-in RouterOS DNS server has no practical per-client query log. Run a separate resolver.
  • IPv6 rules live under /ipv6 firewall. Repeat the redirect and blocks there.
  • UPnP: /ip upnp. While it is on, its mappings appear as dynamic NAT rules.
Aside: Ubiquiti UniFi, for comparison
  • VLANs are Networks. Recent UniFi Network versions use a zone-based firewall, and one policy can cover IPv4 and IPv6. RouterOS keeps two separate rule tables.
  • DNS redirect: a destination NAT rule in the gateway's NAT settings. Older versions lack this in the interface.
  • Like RouterOS, the gateway gives no practical per-client DNS query log. Run a separate resolver.
  • The gateway's mDNS proxy repeats discovery across VLANs. It changes what devices can find, so record its setting. Turn it off for the iot and bench networks during an audit.
  • UPnP is a gateway setting, off by default.

Phase 2

Capture

Capture the trunk between the router's LAN side and the core switch. This is before NAT, so every flow keeps its device MAC and address. Use a mirror port or one inline tap.

  • Blind spot: unicast between two devices on the same switch or access point. Cover it with per-port mirroring for a suspect device, port and client isolation on IoT segments, and canaries.
  • Isolation has a cost: it also breaks casting and local control. Test before you rely on it.
  • Inline taps and PoE: many copper taps do not pass PoE. Mirror the switch port for access points and cameras instead.
  • No mirror-capable gear: export flows (NetFlow or IPFIX) from the router. You get destinations and byte counts, but no payload, no TLS fields and no layer-2 discovery.

Sensor capture interfaces carry no IP address, emit nothing, and keep the VLAN tag:

ip link set eth1 promisc on arp off up
sysctl -w net.ipv6.conf.eth1.disable_ipv6=1
ethtool -K eth1 gro off lro off gso off tso off rx off tx off rxvlan off

Capture into a ring. A new file starts every 300 s or every 1 GB, whichever comes first:

dumpcap -i eth1 -i eth2 -w /data/pcap/trunk.pcapng \
  -b duration:300 -b filesize:1000000 -b files:2000
  • Check interface and dumpcap drop counters daily for the first week.
  • Keep full packets only for iot, lab and bench. For trusted and guest, keep Zeek metadata and delete the packets.

How much disk the packet ring needs

Disk in GB = average Mb/s × days × 10.8

378 GBof packet storage
Average3 days7 days14 days
1 Mb/s32 GB76 GB151 GB
5 Mb/s162 GB378 GB756 GB
20 Mb/s648 GB1.5 TB3.0 TB
50 Mb/s1.6 TB3.8 TB7.6 TB
The average is over the whole day, not the peak. Measure it for one day before choosing retention. Cameras and streaming devices dominate it.

Done whenA test flow from a known device appears in the capture with the right MAC and VLAN tag, in both directions, over IPv4 and IPv6.

Aside: MikroTik RouterOS
  • Port mirroring is a switch-chip feature under /interface ethernet switch. Mirror the router trunk port to the sensor port. Property names differ by RouterOS version and switch chip.
  • /tool sniffer can stream packets to the sensor as TZSP, and /ip traffic-flow exports NetFlow or IPFIX. Both see only packets that reach the CPU. Hardware-offloaded switching, and layer-3 hardware offload on CRS models, bypass them. Use the mirror for evidence and these for quick looks.
  • Port isolation: switch port isolation, or bridge horizon.
  • Not checked on real devices yet.
Aside: Ubiquiti UniFi, for comparison
  • Port mirroring is a per-port setting on UniFi switches. Mirror the gateway uplink port to the sensor port. Check whether your gateway model can mirror its own built-in ports. If not, mirror on the first switch.
  • Quick looks: the Flows and traffic-identification views list per-client destinations, and gateways and access points can run tcpdump over SSH. Recent versions can also export flows as NetFlow or IPFIX. The built-in views use the vendor's own classification, so use the mirror for evidence.
  • The same CPU limit applies: traffic switched in hardware does not reach the gateway's capture tools.
  • Port isolation is a per-port switch setting. Network-level isolation is one checkbox per network.
  • Main difference from RouterOS: fewer capture controls, but less to configure. RouterOS can stream sniffed packets to the sensor (TZSP). UniFi has no equivalent in the interface.

Phase 2, wireless

Wi-Fi

  • The access point's wired uplink is the tap. All client traffic to the gateway crosses it as plain Ethernet.
  • Client-to-client traffic on one access point never reaches the wire. Turn on client isolation for the IoT SSID.
  • Do not plan to read Wi-Fi payload from the air. WPA3 cannot be decrypted from a capture even with the passphrase. The 6 GHz band, wide channels and multi-link operation need a matching capture radio.
  • Use a monitor-mode adapter for management frames only: probe requests and scans. Most IoT clients are on 2.4 GHz.
  • Access point features that proxy, filter or convert multicast change what the wire shows. Record their settings in the inventory.

Done whenA Wi-Fi test device's flows appear on the trunk capture, and a second Wi-Fi device cannot reach it directly.

Aside: TP-Link Wi-Fi 7 PoE access points
  • They are PoE-powered. Mirror their switch ports. No inline tap on those links unless it passes PoE.
  • One IoT SSID mapped to the iot VLAN, client isolation on, 2.4 GHz and 5 GHz only, multi-link operation off for that SSID.
  • 6 GHz requires WPA3, so nothing on that band can be decrypted from the air.
  • Access point management goes on mgmt. If a controller or cloud service manages them, audit that traffic too.
Aside: Ubiquiti UniFi, for comparison
  • UniFi Wi-Fi 7 access points are also PoE-powered, some at higher PoE classes. The same rule applies: mirror the switch port, no inline tap.
  • Per SSID you can choose the bands, turn multi-link operation on or off, and turn on client device isolation.
  • Multicast enhancement converts multicast to unicast, and multicast and broadcast control filters it. Both change what a wired capture shows. Record them, and turn them off on the audited SSID.
  • Access points allow SSH, so a short tcpdump on the access point can show client-to-client traffic that never reaches the wire.

Phase 3

Canaries

Canaries are fake devices in each VLAN. Anything that contacts one is enumerating the network.

  • mDNS adverts for AirPlay, Cast, a printer, SSH, VNC and remote-desktop services.
  • An SSDP responder for a few UPnP device types.
  • Listening services with OpenCanary: SSH, VNC, RDP, HTTP, Telnet.
  • A fake smart plug on TCP and UDP 9999. The video's "system info" requests were probably this plug protocol. Confirm on first capture.
  • Log every query and connection with source MAC and time.

Token rule. Every canary name, MAC, serial and model string contains a unique random token, such as printer-k7q2m9xw. One token per canary per VLAN. A later sighting of a token in decrypted traffic, a vendor privacy export or an ad-profile page is proof of exfiltration.

Done whenA port scan and an mDNS browse from a laptop both show up in the canary log.

Phase 4

Analysis

Zeek turns packets into logs. Load MAC and VLAN logging so every connection carries its device and segment:

zeek -C -r trunk_00001.pcapng \
  policy/protocols/conn/mac-logging policy/protocols/conn/vlan-logging \
  LogAscii::use_json=T

Keep the logs long-term and the packets for days. Query the logs directly:

SELECT orig_l2_addr, "id.resp_h", sum(orig_bytes) AS up
FROM read_json_auto('conn*.log')
GROUP BY 1, 2 ORDER BY up DESC;

Standing queries per device

  • Destinations by domain, network and country. First or third party.
  • Upload bytes per day and per destination.
  • Beacon periods per destination.
  • Discovery: mDNS, SSDP, reverse-lookup and ARP sweeps.

 

  • Peer matrix: which local devices it contacts directly.
  • TLS versions, server names, fingerprints, plaintext protocols, QUIC share.
  • DNS: hard-coded resolvers, encrypted-DNS attempts.
  • Every rate as a count and a window.

Done whenThe standing queries run from one command and produce a per-device table.

Phase 5

Active checks

Run from each VLAN, monthly and after firmware updates:

nmap -sS -sU --top-ports 200 -sV -oA scan-v4 <target>
nmap -6 -sS -sV -oA scan-v6 <target-v6>
  • Record open ports and banners. Fetch UPnP device descriptions.
  • Check firmware versions against known vulnerabilities.
  • Test segmentation: confirm that what the firewall should block is blocked.
  • Audit the network gear too. Look for cloud management, update checks, vendor discovery broadcasts and management services open on user VLANs.
Aside: MikroTik gear as a subject
  • MikroTik devices send neighbour discovery (MNDP, UDP 5678) on their interfaces by default. Expect it in captures. Limit it in /ip neighbor discovery-settings.
  • Check /ip cloud, /ip service, the MAC server and RoMON settings, and record what each device contacts.
Aside: Ubiquiti UniFi, for comparison
  • UniFi gear talks more by default than RouterOS gear. Expect device discovery broadcasts (UDP 10001), inform traffic from every device to the controller (TCP 8080), STUN, firmware checks, and cloud traffic when remote access is on.
  • Record the remote access and analytics settings, then capture what the controller and gateway still contact with both turned off.
  • A controller that manages everything is itself a high-value device. Put it on mgmt and give it a scorecard.

Tier 2  one device at a time

Bench

The bench is the video's rig with the gaps closed. Use it for devices tier 1 flags and for new purchases before they join the network. Bench first the devices that score highest:

  • Has a microphone or camera.
  • The vendor earns from advertising or data.
  • Closed firmware with no local-only mode.
  • Shares a VLAN with personal devices.
  • Tier 1 shows discovery sweeps, peer probing or canary hits.

Run sheet: seven device states under three network conditions

Capture hours per cell

Device stateReal internetFake internetNo answer
Out of box, terms not accepted2 h2 h2 h
Terms accepted, all opt-outs2 h2 h2 h
Terms accepted, all opt-ins2 h2 h2 h
Standby24 h24 h24 h
Active use2 h2 h2 h
External input playing2 h2 h2 h
After firmware update2 h2 h2 h
Column total36 h36 h36 h
The full sheet is 108 hours of capture per device. Run the real-internet column first: it is the one the video lacks. "External input playing" tests content recognition on a TV. Factory reset between rows where a state cannot be undone.

Getting plaintext, in order of preference

  1. Keys from the device.On a rooted device, log TLS keys or install the bench CA in the trust store.
  2. A proxy the device accepts.Record which connections accept a forged certificate. Each one is a finding.
  3. Metadata only.Server names, fingerprints, sizes and timing, as in the table above.

Proving exfiltration

  • Token search. Search decrypted payloads for canary tokens and the device's own identifiers.
  • A/B runs. Same state, canaries present against canaries absent, at least 3 runs per arm. A difference in upload bytes larger than the run-to-run spread points to the destination that receives LAN data.
  • Vendor side. After one week, check the vendor's privacy export and ad-settings pages for tokens.
  • Radio. Monitor-mode Wi-Fi for scans, an nRF52840 dongle for BLE, Zigbee and Thread.

What each audit produces

One scorecard per device. The example below fills the scorecard from the video alone. The open rows are what the extra conditions, tokens and radio capture exist to answer.

Scorecard filled from the video: LG OLED TV, fake internet only

Discovery
Reverse-lookup sweep every 300 s. 57 mDNS service queries in 4 hours.
Measured
Peer probing
99 system-info requests to a fake plug, 54 version requests to a fake bulb, 18 searches for an absent brand.
Measured
TLS hygiene
3,178 handshakes abandoned at the certificate check. 321 sessions completed.
Measured
Destinations
Unknown. The TV never reached a real server.
Open
Upload bytes per state
Not measured. One state, one condition.
Open
Token sightings
Not tested. Canary names were not unique tokens.
Open
Radio
Not captured. The Wi-Fi scan claim stayed unconfirmed.
Open
Verdict
Collection confirmed. Exfiltration not shown either way.
A full scorecard also lists beacon periods, DNS behaviour, behaviour when blocked, open ports, and the allowlist.

From audit to policy

Turn the observed destinations and ports into a per-device firewall allowlist. This is the idea behind Manufacturer Usage Description (RFC 8520). Alert on any flow outside the list, and re-audit after each firmware update.

Lab hosts

On servers, VMs and containers you control, the host tells you more than the wire: per-process egress attribution, TLS key logs, and container egress through one logged gateway. Keep wire capture for opaque boxes such as management controllers, NAS units, cameras and recorders.

Privacy and scope

  • Only devices and networks you own.
  • Full packet capture only on iot, lab and bench.
  • Metadata only for household members' and guests' personal devices. Tell them it exists.
  • Proxy interception only on bench.
  • Captures and key logs hold account tokens and passwords. Treat them as secrets and keep them out of version control.

Order of work

  1. Settle the open decisions and rank devices.Tier 1
  2. Inventory, segmentation and DNS. Resolver logs alone already show reverse-lookup sweeps.Tier 1
  3. Canaries on the existing network. Cheap, immediate signal.Tier 1
  4. Sensor, Wi-Fi checks and analysis.Tier 1
  5. Active checks, including the network gear.Tier 1
  6. Bench the first subject.Tier 2
  7. Scorecards and allowlists, device by device.Tier 2

Open decisions

  • Router and firewall: VLANs, per-VLAN rules, DNS redirect, UPnP mappings visible?
  • Core switch: port mirroring? If not, one inline tap on the trunk, or flow export.
  • Access points: a VLAN per SSID, and client isolation?
  • Sensor: a small x86 box with two or more wired NICs and about 1 TB of fast storage. A single-board computer with USB NICs suits the bench, not the trunk.
  • Which devices are rooted or can be? They are the ones that give plaintext.
  • First bench subject.