WISP Signal Troubleshooting for Support Agents
One slow subscriber or a whole sector? Use signal, noise, link quality and traffic evidence to choose what to check first, before dispatching a technician.
Read the whole radio link
Start with the scope of the fault
Ask when the problem started, whether every device is affected, and whether nearby subscribers have called. Record the subscriber, CPE, AP radio/interface, timestamp and timezone. An SSID alone does not identify a sector. A cluster of calls suggests a shared investigation, not a proven common cause.
Use this path to separate the customer's indoor Wi-Fi from the outdoor access link:
Customer device → indoor Wi-Fi / LAN → outdoor CPE → AP sector
→ Ethernet uplink → backhaul → upstream
Repeat the customer's test over Ethernet where practical, then compare loss, latency and throughput at the access gateway and farther upstream. A good radio reading does not prove working internet. If several sectors degrade together, inspect their shared uplink or backhaul before realigning individual CPEs. Keep these observations with your WISP operations workflow.
What the radio numbers mean
Read both ends in the same time window, with known traffic activity. Save the radio model, firmware, band, frequency, channel width and per-chain readings alongside the numbers.
| Metric | Meaning | First question |
|---|---|---|
| RSSI / signal strength, dBm | Received power. -60 dBm is stronger than -75 dBm. | Did it drop from this link's normal level, or does one antenna chain differ? |
| Noise floor, dBm | The receiver's estimate of background RF energy. -85 dBm is noisier than -95 dBm. | Did noise rise on this same receiver, frequency and width? |
| SNR, dB | Signal above noise: signal dBm minus noise dBm. | Is there enough margin for the expected modulation? |
| TX/RX CCQ, % | MikroTik's connection-efficiency metric, affected by rate and retries. | Is efficiency low even when received power looks normal? |
| airMAX quality / capacity | Separate vendor indicators of link quality and airtime efficiency. | Is quality falling, or are slow clients consuming airtime? |
| TX/RX rate, Mbps | Radio PHY rate in each direction, not customer download speed. | Does an active link reach its known healthy rate for this configuration? |
MikroTik documents the legacy fields in its wireless registration table and explains CCQ in Wireless Troubleshooting. The newer WiFi registration table exposes a different field set: do not assume legacy CCQ exists or treat traffic bits per second as PHY rate. Package and firmware matter more than the version number alone.
Ubiquiti's airOS 6 User Guide defines noise, CCQ, rates, airMAX quality (AMQ) and capacity (AMC) for that product generation. AMQ reflects retries and physical link quality; AMC reflects airtime efficiency. Its advice that AMQ above 80% needs no adjustment when there are no other symptoms is specific to that guide, not a universal CCQ threshold or proof of health. Verify the manual for the installed model and firmware. These vendor references describe manual observations, not ISPAgents integration support.
Good, marginal and poor: initial triage ranges
These are ISPAgents triage rules of thumb, not vendor specifications, guaranteed speeds or installation acceptance targets. They match the proposed subscriber-link diagnostic policy. Validate them against the link budget and healthy busy-hour baseline for each radio model. Stronger signal can still exceed a receiver's recommended input level; the first row only checks for weakness.
| Metric | Good metric / no flag | Marginal / degraded | Poor |
|---|---|---|---|
| Weakest measured signal or antenna chain | -67 dBm or stronger | -75 dBm to below -67 dBm | Below -75 dBm |
| SNR | 25 dB or more | 15 dB to below 25 dB | Below 15 dB |
| Either TX or RX CCQ | 85% or more | 60% to below 85% | Below 60% |
| Active TX or RX PHY rate / healthy expected PHY rate | 50% or more | 25% to below 50% | Below 25% |
A good single metric cannot clear the entire link. Missing, stale or unsupported measurements remain unknown. Prefer the worst observed problem for investigation, then verify the cause. Low SNR can reflect weak signal, high noise, or both.
For example, -63 dBm signal minus -92 dBm noise gives 29 dB SNR. With the same signal and -78 dBm noise, SNR is 15 dB, at the marginal boundary. Use signal and noise from the same receiver; do not subtract AP noise from a CPE-side signal reading.
For rates, a measured 30 Mbps against a healthy 130 Mbps PHY baseline is about 23%, in the poor band. A 10 Mbps subscription is not that baseline. Compare matching band, width, radio standard, streams and guard interval, in each direction while carrying traffic. An idle radio may show only its last low transmit rate. Without activity or a valid baseline, leave rate assessment unknown.
Noise has no universal good/bad scale across radios and widths. As an advisory sector heuristic, investigate AP noise at -85 dBm or higher, or a rise of at least 10 dB against a comparable recent reading. A lower value does not clear interference; correlate SNR, retries and spectrum evidence. Compare the same receiver, calibration, frequency and width; the planned sector policy uses prior observations within 15 minutes.
One subscriber link, a busy sector, or interference?
| Evidence pattern | Likely investigation | What to check next |
|---|---|---|
| One CPE has weaker signal; peers are stable | Subscriber path, alignment or hardware | Per-chain levels, mount movement, obstructions, PoE and Ethernet errors |
| Many peers have stable RSSI but lower SNR / higher noise | Shared RF interference | Noise history and a planned spectrum survey at both ends |
| Busy-hour slowdown, high airtime, mostly stable RF | Sector load or inefficient clients | Active-client rates, airtime, traffic and queues; compare off-peak |
| Low radio rates from one active weak client | Possible disproportionate airtime use | Verify actual airtime evidence before attributing other users' faults to that client |
| Radio evidence stable, several sectors slow | Shared uplink, backhaul or upstream | Ethernet errors, negotiated speed, capacity, queues and loss along the path |
| Wired test works, only indoor wireless is slow | Customer Wi-Fi / LAN | Indoor coverage, contention and client capability |
For a sector, collect a complete registration table and the affected subscribers' timestamps. Compare the median and weakest tenth (p10), as well as individual outliers. Unknown metrics are not healthy readings. A low-SNR subset alone does not establish sector-wide interference.
The planned sector diagnostic uses 85% measured airtime as a saturation flag and 20% or more below-good links as a degradation flag. Its default client-count flags are more than 40 clients at 20 MHz, 60 at 40 MHz or 80 at 80 MHz. These are configurable investigation thresholds, not vendor capacities or recommended design limits. Exactly at the cap does not trigger the count flag; it still does not prove spare capacity. Traffic mix, protocol and radio generation change usable capacity. Increasing channel width just to raise the count threshold is not a fix.
What to fix first
- Preserve evidence and rule out power or Ethernet trouble. Record uptime changes, link negotiation, cable errors and PoE condition. Compare a wired test with the customer's Wi-Fi test before dispatch.
- Check alignment and the path for an isolated weak link. Compare both antenna chains and both directions against the installation baseline. Inspect mount movement, polarization, connectors and new obstructions. Do not immediately raise transmit power to hide a path problem.
- Check line of sight and Fresnel clearance. Seeing the other antenna is not enough: the radio path occupies space around the direct line. As a planning rule of thumb, keep at least 60% of the first Fresnel zone clear, then validate the full path profile, vegetation growth, terrain and antenna heights for the actual frequency and distance. The Cisco site-planning guide explains why visual line of sight and Fresnel clearance differ.
- Investigate shared noise before changing a sector. Survey both ends with the device's supported tools. Ubiquiti warns that airView on airMAX M interrupts wireless service; schedule that check and retain an access/recovery path. Do not assume another radio generation behaves identically.
- Review channel width and load together. A narrower channel can avoid some interference but lowers potential PHY capacity; a wider channel consumes more spectrum. Check both ends' support and the permitted channel plan, record the original settings, then change one variable in an approved window. Retest latency, loss and wired throughput under comparable traffic and revert if service worsens.
Printable field checklist
Open the printable WISP signal checklist, then use your browser's Print command or save it as PDF. It is a static worksheet with no submission or data collection.
- Identify the customer, CPE and exact AP radio; record time, symptoms and affected peers.
- Capture model/firmware, frequency, width, both-end signal/chains, noise, SNR, CCQ or named vendor quality metric, TX/RX PHY rates and activity.
- Record wired versus Wi-Fi results, power/Ethernet checks, path clearance and busy-hour comparison.
- Agree one change, a recovery path and a retest; keep before/after evidence and the next owner.
Where ISPAgents fits today
Available: basic subscriber/device association and TR-069/CWMP diagnostics provide a starting point for operator evidence, within the supported device scope. MikroTik RouterOS REST operations and SNMP monitoring are guided pilot capabilities with model, firmware and release limits.
Wireless subscriber-link and sector verdicts in the product UI are planned. This guide does not promise automatic RF diagnosis, airMAX telemetry ingestion or automatic channel changes. Discuss a guided pilot against your actual equipment and retain your existing radio tools. See the WISP management scope or return to the guides index.
Frequently asked questions
What is a good WISP signal strength?
For initial triage, -67 dBm or stronger clears the weak-signal check; -75 to below -67 dBm is marginal, and below -75 dBm is poor. These are heuristics, not installation targets. Use the radio model's link budget, receiver limits, SNR and active-traffic evidence before calling a link healthy.
Can strong RSSI still give slow internet?
Yes. Interference can reduce SNR, retries can consume airtime, or the sector, Ethernet uplink, backhaul or upstream can be congested. Compare other customers on the same AP radio and repeat the test over Ethernet to separate access-radio trouble from indoor Wi-Fi.
Are CCQ and airMAX quality the same?
No. Both can help investigate link efficiency, but they use different vendor definitions. Keep the original metric name and use its model-specific documentation. Do not turn airMAX quality, capacity or retries into a fabricated CCQ value.
Does ISPAgents show these wireless verdicts today?
Wireless subscriber-link and sector verdicts in the product UI are planned. SNMP and MikroTik RouterOS REST operations are guided pilot capabilities with model, firmware and release limits; this guide does not establish airMAX telemetry coverage.
Continue the operations map.
One Subscriber, One Device Identity
Available: The canonical identity foundation supports mapping protocol observations toward one device/customer path using strong serial/MAC evidence. It does not establish universal billing, RADIUS, or controller integration.
Open pageGuidesFair-usage policies for unlimited ISP plans
Keep access available after a usage threshold, explain the reduced speed before purchase, and give support a policy they can defend.
Open pageSolutionsAutomatic Internet Suspension Software — Guided pilot
Guided pilot: Preview, queued jobs, approvals, result recording, and a mock-safe worker can be evaluated in a guided pilot. Live router/RADIUS mutation remains gated; automated suspension and restoration are not a shipped promise.
Open pageSolutionsCustomer Self-Service Foundations — Guided pilot
Guided pilot: Field-job lifecycle, serialized stock, basic tickets and subscriber portal flows have narrow synthetic-data proof and remain guided pilots. Physical installs and invite email delivery are unproven. The technician PWA, dispatch board, dedicated stock workspace and IPAM remain roadmap.
Open page