What Is Interfering With Your Wi-Fi at Home (and How to Prove It)

The call freezes for ninety seconds and comes back on its own. The hallway smart plug drops every evening at about the same time. Your phone shows three bars throughout. That combination — strong signal, unreliable behavior, a pattern in time — is interference, and it is the one Wi-Fi problem that a stronger signal does not fix.

Here is what is actually emitting in your house, the signature each one leaves, and a test sequence that names the culprit in an afternoon.

Interference does not weaken your signal#

Your radio needs the signal to stand out from the background, not merely to be loud. Interference leaves the signal exactly where it was and lifts the background underneath it.

That gap between the two is the signal-to-noise ratio. Above about 25 dB of margin everything a home does works; 20 dB is the practical minimum for full throughput. Below 15 dB the link falls apart however strong the signal reading looks, which is why a phone can sit at a healthy -55 dBm and still stall. Those thresholds sit alongside the dBm and SNR reference table.

One question separates interference from a coverage problem:

• Coverage problems have a map. They fail in the same place at every hour. Walk ten feet and they improve.
• Interference problems have a clock. They fail at a time — while something is running, while someone is home, during the reheat cycle — and they are fine in the same spot an hour later.

If yours has a map rather than a clock, start with how dead zones are built instead. Plenty of rooms have both; run the interference test first, because it is the free one.

Two kinds of interference: radios that take turns, and radios that do not#

Everything that harms your Wi-Fi falls into one of two families.

Other Wi-Fi. Wi-Fi radios listen before they transmit. If another network is on your exact channel, the two hear each other and take turns. That is co-channel interference: you keep a clean link and lose airtime, so throughput drops and latency gets lumpy. If the other network only partly overlaps you — they are on channel 3, you are on 1 — that is adjacent-channel interference. Their energy lands inside your channel but never decodes as a Wi-Fi preamble, so neither radio defers and frames get corrupted mid-flight. Partial overlap is worse than sharing. Choosing channels and widths is a job of its own, covered in channels, neighbors and noise.

Everything else. A microwave oven, a video sender or a badly shielded cable does not speak Wi-Fi at all. Nothing defers to it and your router cannot negotiate with it; its energy simply raises the noise floor for as long as it transmits.

The suspects, and the signature each one leaves#

Almost every household source lives in 2.4 GHz. The few that do not are at the bottom of the table.

If your problem is centered on the kitchen counter, go to microwaves, metal and 2.4 GHz.

Duty cycle matters more than power#

Rank suspects by how much of each second they occupy, not by how loud they are. A Zigbee door sensor transmits a few milliseconds an hour and is harmless. An analog video monitor transmits a continuous carrier and is catastrophic at a fraction of the power, because it leaves no gap for your frames.

Duty cycle also explains a common misdiagnosis. A Zigbee or Thread mesh is almost always the victim, not the aggressor: those radios run about a milliwatt against a router running a hundred times that. Dropping sensors are usually at the edge of coverage on a 2.4 GHz-only radio — a different failure with a different fix.

The band swap is your best single diagnostic#

Nearly everything in that table is a 2.4 GHz problem, so the fastest test here is to move the struggling device onto 5 GHz and repeat whatever failed. If the failure disappears you have eliminated ten suspects at once, and you already know the fix: keep laptops, phones and TVs on 5 GHz and leave 2.4 GHz to the sensors, the argument made in full in 2.4 vs 5 vs 6 GHz.

The isolation test, step by step#

You need one thing you can watch continuously and one rule: change one variable at a time.

1. Set up a live indicator. A continuous ping to your router's LAN address is ideal: you get average latency and packet loss. A video call with a visible quality indicator works too. A speed test does not: it samples 20 seconds and misses anything that fires in bursts.
2. Record a baseline. Sixty seconds in the seat where the problem happens, on the band that fails, nothing changed. Note average latency, the worst spike, and loss.
3. Run the band control. Join 5 GHz — give it its own name temporarily if band steering keeps moving you — and repeat the baseline. Clean on 5 GHz and bad on 2.4 GHz puts your suspect in the table above. Equally bad on both narrows it to neighbors, a video sender, DFS radar, or something that is not interference.
4. Toggle one suspect. Power it off, wait 60 seconds, measure. Power it on, wait 60 seconds, measure. Twice per suspect: one convincing trial is a coincidence, two matching pairs are evidence.
5. Walk fifteen feet away and repeat the worst case. Most household emitters are short-range. If the trouble follows you across the house, suspect the neighbors or your own channel choice, not a gadget on the desk.
6. Repeat at the hour it hurts, plus one quiet hour. A problem that opens at 8 p.m. and is gone at 6 a.m. is the building, not your kitchen.
7. Log four columns: time, what changed, average latency, loss. Six rows beat any amount of recollection.

Expect the numbers to move on their own. Signal strength wanders 3 to 5 dB while you stand still and latency wobbles with it, so only a swing beyond about 10 dB, or a clear jump in packet loss, counts as a result.

Usually blamed, usually innocent#

Do not spend a weekend on these.

• Modern cordless phones. DECT handsets in the US sit near 1.9 GHz, nowhere near Wi-Fi. Only older 2.4 GHz and 5.8 GHz models are suspects.
• LED bulbs, dimmers, fluorescent tubes. Cheap drivers make electrical noise, but it lives far below the Wi-Fi bands. They ruin AM radio, not your video call.
• The fridge, the TV, the water heater. Not emitters. They are metal, and metal is a coverage problem.
• Bluetooth headphones at normal distance. Bluetooth hops away from busy channels at a tiny fraction of a router's power. It matters when two antennas are inches apart, rarely otherwise.
• A neighbor's 5 GHz network two floors down. It arrives close to the noise floor. Their 2.4 GHz network is the one competing with you.

When an extender is the wrong tool for this#

This is the part worth being blunt about: an extender does not reduce noise, it adds to it.

A repeater is one more radio on your channel, in your house. On a single-radio unit every packet is sent twice, so the airtime each device behind it consumes roughly doubles. Put one in a room whose real problem is a noisy channel and you have added a second loud voice to a room that was already too loud — plus a strong-looking access point for your phone to cling to.

An extender earns its place only when the room genuinely has weak signal. If you get there, put it where the router still reads -60 to -67 dBm, following the placement rule that works.

Where planning still helps#

Range Up cannot see any of this. It does not listen to the air, does not read SSIDs and does not run speed tests — it models the building. That sounds useless for a noise problem until you notice what a diagnosis like this is missing: a control. Every toggle test above compares the room against itself, with no independent statement of what the room ought to be getting.

That is the number a simulation supplies. Outline the floor, set the material preset to what the building is made of, put the router on its real shelf, and the app returns a predicted level for the seat that keeps freezing. Then the test above becomes a two-way decision. Predicted -60 dBm or better and the call still breaks up? Coverage is not your problem, no box fixes it, and the culprit is in the suspects table. Predicted down in the -70s and the room was starved all along, which turns this into a placement question — mark that seat as the room that matters and let the app rank the outlets you could plug into. The extender placement guide takes it from there, and the app itself is a free download.