Finding Rogue Access Points
Of all the things people plug into a wall jack without asking, the wireless access point is the one that keeps network administrators up at night. A desk switch extends the network by a few feet of cable. A personal access point extends it to the parking lot. Whoever can reach its radio can reach your LAN, subject only to whatever password the person who bought it remembered to set. Every access control your managed wireless enforces — authentication, guest isolation, logging — is bypassed in one stroke.
The good news is that a rogue AP is unusually loud. It leaves footprints on two sides at once. On the wired side, it must connect to a switch port to be useful. On the wireless side, it must broadcast to be found by its owner's phone. This guide teaches you to read both sets of footprints. It teaches you to correlate them so you do not go to war with the coffee shop next door. You will learn to walk a building and localize a radio. And you will learn to contain a confirmed rogue without breaking anyone else's network or the law.
In this guide
- Two ways an AP plugs in: router mode and bridge mode
- Wired-side signals
- Wireless-side signals
- The neighbor problem: correlate before you act
- Walking with a scanner to localize it
- Wired versus wireless evidence at a glance
- Containment cautions
- An annotated example: the conference-room router
- Where to go next
Two ways an AP plugs in: router mode and bridge mode
Before the signals make sense, you need one distinction. A consumer wireless box can attach to your network in two modes, and they look completely different from the wired side.
In router mode — the factory default for nearly every home router — the box's "internet" port goes into your wall jack. Everything on its wireless side sits behind NAT (network address translation). Your switch sees exactly one MAC address and one IP: the router's. A dozen phones behind it are invisible to you, all sharing that single identity. The router also, by default, runs its own DHCP server on its wireless side. That is harmless while it stays there. It is a floor-wide outage if someone plugs the wrong port into the wall. That failure mode is the subject of Rogue DHCP Servers.
In bridge mode (or "access point mode"), the box simply extends your LAN over radio. Every wireless client gets its own MAC on your switch port and its own address from your DHCP server. This is the mode a slightly technical person chooses, and it is louder on the wired side: the port suddenly hosts many MACs. Router mode is quieter on the wire but gives itself away in other ways. Learn to recognize both.
Wired-side signals
The wired side is where you have the most control and the most reliable data, so start here. Everything below comes from switch MAC tables, DHCP logs, and traffic counters you probably already collect; New-MAC Alerting covers how to pull them.
Many MACs on one access port
An access port serving a desk should host one MAC, or two if there is a phone with a PC behind it. A port that suddenly hosts five, eight, or fifteen is hosting either a switch or a bridged AP. The AP case has a tell. The MACs come and go through the day as people walk in and out of range. Several are likely to be randomized (phones on an unfamiliar network randomize by default). A desk switch's MACs, by contrast, are stable, and mostly belong to desktops and printers. Counting MACs per port from the table you already poll is the cheapest AP detector you will ever build.
One MAC that is secretly many devices
Router mode hides the crowd, but not perfectly. The single MAC will belong to a vendor prefix associated with consumer networking gear. Its DHCP request will very often carry a hostname straight from the factory — a model number, or a brand followed by "router." And its traffic looks wrong for one device. Suppose a "desktop" maintains connections to several video-streaming services, three different phone-vendor update servers, and a games console's matchmaking network at the same time. That is not a desktop. If you export flow records from your firewall, the number of simultaneous distinct destinations from one internal address is a strong hint. So is the packet TTL (time-to-live, a hop counter every router decrements). Traffic from behind a NAT router arrives with a TTL one lower than a directly connected machine would send. A firewall or a capture can show you that.
DHCP fingerprints
Whichever mode the box is in, something asks your DHCP server for an address: the router itself in router mode, or each wireless client in bridge mode. The lease log then shows either one consumer-router hostname or a sudden cluster of phone-like hostnames all appearing through the same switch port. Many DHCP servers also log the vendor class the client sends, which for phones and consumer gear is often quite specific.
Traffic patterns from an access port
A desk port has a daily rhythm: quiet overnight, busy during working hours, a lunch dip. A port with an AP behind it has a different shape. It has traffic at odd hours (a phone updating at midnight) and bursts that do not match the desk's occupant. And its total volume is high for "one PC." If you graph per-port traffic, an access port whose weekly shape looks like a wireless network's rather than a desktop's is worth a look. The port counters that make this possible are described in Interface Metrics That Matter.
Port security and 802.1X events
If your access ports limit MACs or require authentication, a bridged AP trips the limit almost immediately. Under those rules, a router-mode AP fails authentication unless someone has gone to real effort. The switch logs the violation with the port name. That is the shortest path from "AP plugged in" to "alert with location," and it is covered in Port Security and 802.1X as Detection Signals.
Wireless-side signals
Your managed access points are also radios, and radios hear everything on their channels. Nearly every wireless controller or management system — on-premises or cloud-managed — has some form of rogue detection. The APs periodically scan. They note every network name (SSID) and radio identifier (BSSID, the MAC of the transmitting radio) they hear. They report the ones that are not part of your own deployment. The feature goes by different names, but the reports contain the same fields, and three of them matter most.
The SSID. A network named after a person, a brand, or a model number is probably a personal AP. A network named exactly like yours, or one letter off, is a different and more serious animal. It may be a neighbor's coincidence, or it may be an attempt to lure your users into connecting to the wrong network. Treat lookalike SSIDs as a priority.
The BSSID. Like any MAC, it carries a vendor prefix — a hint, as always, not proof. More usefully, consumer devices very often derive their wireless BSSIDs from their wired MAC by changing only the last digit or two. Suppose a managed AP reports a BSSID of 00:1a:2b:3c:4d:5f and your switch table has 00:1a:2b:3c:4d:5e on a desk port. Then you have very likely found both ends of the same box. Many controllers automate exactly this correlation and label the result "rogue on wire" or similar.
Signal strength. Each of your APs reports how loudly it hears the rogue. Suppose three of your APs hear it. If the one on the second floor east side hears it far more strongly than the others, the rogue is near that area. This is coarse — walls, floors, and antenna orientation confuse it. But it usually narrows the search to a wing, which is what you need before you start walking.
Tip: get a baseline of the wireless-side report on a quiet day. Note every neighboring SSID and BSSID your APs routinely hear and record them as known neighbors. Once you have that list, the rogue report becomes a diff against a baseline rather than a wall of unfamiliar names. A genuinely new radio stands out. The same "compare against what you know" habit that makes new-MAC alerting work makes rogue reports readable.
The neighbor problem: correlate before you act
Here is the most important idea in this guide: a wireless network your APs can hear is not necessarily on your network. Radio does not respect property lines. The coffee shop downstairs, the law office across the hall, and the apartment building next door will all appear in your rogue report. So will every phone hotspot in your lobby. None of them are your problem. A controller that labels them all "rogue" is using the word loosely. The distinction you need is:
- Neighbor: a radio you can hear that is connected to somebody else's wired network. Not your rogue. Not your business.
- Rogue on your wire: a radio you can hear and that is bridged into or routed through one of your switch ports. This is the one you act on.
- Lookalike: a radio broadcasting your SSID (or a close imitation) regardless of where it is wired. Worth investigating even if it is not on your wire, because it targets your users.
Moving a report from "neighbor" to "rogue on your wire" requires wired-side evidence. That includes the BSSID-adjacent MAC on a switch port, the consumer-router hostname in your DHCP log, the many-MACs port, or the port-security violation. Until you have at least one of those, you have a radio, not a rogue. Some controllers try to establish the wired connection automatically. For example, they watch for MAC addresses seen both over the air and on the LAN. When they say "rogue on wire," trust but verify against your own switch table. The correlation can misfire when a legitimate client roams between wired and wireless.
Walking with a scanner to localize it
Suppose the signal-strength hints have narrowed the search to a wing and the wired evidence has told you which switch port. Then you usually already know the room, because the port label says so. When the port label does not exist or the wired trail runs cold, you walk. This applies only to buildings and networks you are authorized to manage; it is your own premises you are searching.
Any Wi-Fi analyzer app on a laptop or phone will list the SSIDs and BSSIDs it hears with a signal-strength reading. Filter to the BSSID you are hunting and watch the number as you move. Signal strength is reported in dBm, a logarithmic scale where less negative is stronger. A reading around -40 means you are very close. Around -70 means the same room or an adjacent one. And -85 means it is somewhere in the building. Walk toward stronger; when the number peaks and starts dropping again, you have passed it. Then look under desks, on top of cabinets, behind monitors, and in the ceiling void above the strongest spot. Consumer routers have blinking lights, which helps; deliberately hidden devices do not, which is itself information.
A few practicalities. Radios on the 2.4 GHz band pass through walls better than those on the 5 GHz band. So hunt with whichever band the rogue is using, and expect the 2.4 GHz trail to be blurrier. Do the walk during working hours if you can. A router-mode AP with nobody connected is still broadcasting. But a bridged one is easiest to confirm when clients are on it and their MACs are appearing on your port. And bring a colleague if the building is large; two readings from two positions triangulate faster than one person pacing.
Wired versus wireless evidence at a glance
| Evidence | Side | What it suggests | Strength on its own |
|---|---|---|---|
| Many MACs on one access port, several randomized, coming and going | Wired | Bridged AP (or a desk switch with wireless clients behind it) | Strong |
| One MAC with a consumer-networking vendor prefix and a factory hostname | Wired | Router-mode AP | Moderate to strong |
| Cluster of phone-like DHCP hostnames arriving through one port | Wired | Bridged AP | Strong |
| Traffic to many streaming or phone-update destinations from one "desktop" IP; TTL one lower than expected | Wired | Router-mode AP hiding several devices | Moderate |
| Port-security MAC-limit violation or 802.1X failure on a desk port | Wired | Something unexpected plugged in; AP is a common cause | Moderate (location is exact; identity is not) |
| Managed APs report an unknown SSID | Wireless | A radio exists nearby — could be a neighbor | Weak alone |
| Unknown BSSID adjacent to a MAC on your switch table | Both | Rogue on your wire | Strong (this is the correlation you want) |
| SSID identical or near-identical to yours | Wireless | Lookalike; possible attempt to lure your users | Priority regardless of wire status |
| Signal strength much higher at one managed AP than others | Wireless | Location narrowed to a wing | Location only, not identity |
Read the "strength" column as a rule for action. A single wireless-side row justifies investigation. A wired-side row justifies locating the port. The correlation row — both sides pointing at the same box — justifies containment. Acting on the wireless side alone is how teams end up disabling a neighbor's network.
Containment cautions
Many wireless controllers offer containment: your managed APs actively disrupt the rogue's radio so that clients cannot stay connected to it. It sounds like exactly what you want, and it is dangerous for two reasons.
First, radio does not stop at your walls. Containing a radio that turns out to belong to the business next door means you are deliberately interfering with somebody else's legitimate network. That is at minimum a very awkward conversation. And, in many jurisdictions, it may be unlawful regardless of your intentions. Second, containment does not remove the device. It is still on your switch port, still bridged to your LAN, and will resume the moment containment stops or a client finds a gap.
The rule is therefore simple. Verify the rogue is on your wire before any containment, and when it is, prefer the wired fix. Shut down the switch port (or move it to a quarantine VLAN), then go and collect the device. Disabling one port on your own switch affects nobody but the rogue, is instantly reversible, and produces a very motivated owner who will come and find you. Reserve radio containment for the narrow case where a lookalike SSID is actively targeting your users and you cannot yet find its wire. Even then, get a second person to confirm before you enable it, and keep it as short as possible.
One more caution: your managed APs' own health matters here. A controller whose rogue-detection scanning is disabled, or an AP whose radios are so busy serving clients that they never scan, will under-report. If rogue reports seem suspiciously empty, check that the scanning is actually running; AP Health Monitoring covers keeping the managed side healthy enough to watch the unmanaged side.
An annotated example: the conference-room router
A worked case, start to finish, in the order the evidence actually arrived.
- The first signal (wireless): the weekly rogue report lists a new SSID,
Marketing-Fast, heard by three managed APs, strongest on the third floor west. Its BSSID's vendor prefix belongs to a consumer-router maker. On its own: a radio near the third floor west, possibly a neighbor. - The correlation (wired): the switch MAC table for the third-floor access switch has a MAC one digit away from that BSSID, on port Gi1/0/22. The port description says
3W conf room B — wall jack. The DHCP log shows that MAC requesting an address with a factory hostname. This is now a rogue on your wire, in router mode. - The assessment: only one MAC on the port, so it is NATing its clients. Flow records from the firewall show that one IP maintaining connections to two video-conferencing services and a personal cloud-storage provider at once. That means several devices behind it. Its DHCP server is confined to its wireless side; nobody else on the floor is getting bad leases. Low urgency, real risk.
- The decision: not malicious; the SSID practically names the department. No containment. The port is shut at the switch at the end of the day, with the description updated to note why.
- The conversation: the marketing team put it there because the conference room's wireless was slow during all-hands calls. They needed better coverage, not a rogue. A managed AP is added to the room; the router goes home with its owner.
- The record: the BSSID is added to the known-neighbor baseline as "resolved rogue, removed," so its reappearance would alert again. And the case is documented following Rogue Device Response.
Notice that the wireless report started the case but did not close it; the wired side did. Notice also what did not happen. Nobody containment-jammed a radio on the third floor, and nobody accused marketing of anything. The underlying problem — bad coverage in that room — was actually fixed. That is the only outcome that prevents the sequel.
Where to go next
You now know how a rogue AP gives itself away on both sides. On the wire, the clues are MAC counts, vendor prefixes, DHCP hostnames, traffic shape, and port-security events. Over the air, the clues come through your own APs' rogue reports, lookalike SSIDs, adjacent BSSIDs, and signal strength. You know that a heard radio is not a rogue until wired evidence says so. You know how to walk a building with a scanner, and why disabling a switch port beats containment almost every time.
The wired signals here depend on the collection loop in New-MAC Alerting and the switch-side events in Port Security and 802.1X as Detection Signals. The categories for judging what you have found are in What Makes a Device Rogue. Everything in this topic is collected on the Rogue Device Detection pillar page.
Put this into practice with SysPeep Monitor
Most of the wired-side evidence in this guide is data your switches already hold and just need someone to collect. SysPeep Monitor auto-discovers your switches and access points. It polls port status and traffic counters over SNMP. It graphs them so an access port that suddenly behaves like a wireless network stands out. It alerts on the changes you care about. The free trial installs on any Windows machine in minutes.
Download Free TrialLearn About SysPeep