Fast speed, unstable ping
Use the Ping Test and compare quiet and busy periods.
5G broadband can show strong download speed but still feel laggy when latency is unstable. This guide helps you validate whether the cause is signal, mast load, Wi‑Fi, router placement or loaded latency.
5G troubleshooting guide
Fast downloads do not guarantee stable gaming or calls. 5G lag usually comes from changing radio quality, mast congestion, home Wi‑Fi, packet loss or bufferbloat when the connection becomes busy.
Quick-fix checklist
Stop game updates, cloud backups, CCTV uploads and large downloads, then check whether the lag immediately changes.
Test now and again during the normal problem period so a peak-time mast pattern is not mistaken for a permanent fault.
A stable wired result points to the home wireless network; an unstable wired result keeps the 5G radio or provider path in scope.
Try another window, a higher position and a different orientation. Keep the change only when ping, jitter or packet loss improves repeatedly.
Interactive troubleshooter
Select the closest match. The result points to one test and one detailed section, so you do not need to work through every technical check.
Your next step
Fast answer
5G broadband can feel laggy despite fast download speeds because gaming and calls depend on stable latency rather than throughput alone. Mast congestion, changing radio signal, home Wi‑Fi, packet loss and bufferbloat can all make ping rise or fluctuate while an ordinary speed test still appears fast.
Use the Ping Test and compare quiet and busy periods.
Use the Packet Loss Test before changing NAT or DNS settings.
Use the Bufferbloat Test while the 5G connection is otherwise quiet.
Find the fault boundary
Run the same test from the same device over Ethernet and Wi‑Fi. This one comparison prevents a room-specific wireless problem from being mistaken for mast congestion or weak 5G coverage.
The 5G link is probably usable at that moment. Improve router placement, use a cleaner Wi‑Fi band, test closer to the router or connect the affected gaming device by Ethernet.
Keep the radio and provider path in scope. Compare peak and off-peak ping, note signal metrics and repeat the test after moving the 5G router.
Check that device, server region, VPN, game route or NAT behaviour before changing the 5G service. A whole-connection fault normally affects more than one real-time app.
Latency diagnosis
A ping spike is a short increase in delay. On 5G it can come from changing radio conditions, a cell or band change, mast scheduling, packet loss, Wi‑Fi interference or an upload queue filling.
If only the router ping jumps, focus on Wi‑Fi or the local network. If the router stays stable but the external ping jumps, focus on the 5G and provider path.
A consistent evening pattern points more strongly toward mast or provider contention than a random one-off spike.
Freezes with loss need the packet-loss route; spikes that begin only during uploads or downloads need the bufferbloat route.
Radio and mast analysis
Mast congestion is likely when Ethernet remains stable locally but external ping, jitter or packet loss worsens repeatedly during busy periods. Compare the same test off peak and at the normal problem time before blaming the router.
The mast allocates time and frequency resources among active users. As sector load rises, packets may wait longer for transmission opportunities, which increases round-trip-time variation.
Many 5G deployments use time-division duplexing with more radio capacity assigned to downloads than uploads. Cloud backups, cameras and livestreaming can therefore create large uplink queues and delay game inputs.
A router can show strong RSRP while quality and latency deteriorate because of interference, sector load or radio scheduling. Compare RSRQ, SINR and live latency rather than relying on bars alone.
Moving the router to a different window or height can alter the selected band, sector or mast. Small changes may improve SINR and jitter even when download speed changes only slightly.
Use Ethernet for the test where possible so Wi-Fi does not distort the comparison.
Record minimum, average and maximum latency, packet loss, download, upload and the router's signal metrics.
Use the same destination and test length. Also keep a simultaneous ping to the router gateway.
If the gateway remains stable but external latency and loss worsen significantly at peak time, mast or provider-network contention is more likely than home Wi-Fi.
| Metric | What it describes | Useful interpretation | Rough starting guide |
|---|---|---|---|
| RSRP | Reference signal strength | Less negative is stronger. Very weak RSRP can make the radio link unstable. | Better than about −80 dBm is a strong starting point. Investigate repeatable results worse than about −110 dBm. |
| RSRQ | Reference signal quality | A worsening value can indicate interference or increased cell load. | Better than about −10 dB is a useful starting point. Investigate repeatable results worse than about −15 dB. |
| SINR / SNR | Signal compared with noise and interference | Higher is normally better; low or rapidly changing SINR often aligns with unstable throughput and latency. | Above about 20 dB is a strong starting point. Investigate repeatable results below about 5–10 dB. |
| Cell ID / band | Serving cell and radio band | A change after repositioning can explain a sudden improvement or deterioration. | Prefer a stable cell and band during like-for-like tests. A change is evidence to correlate, not a fault by itself. |
Use these as comparison ranges, not universal pass/fail limits. Router reporting, frequency, bandwidth, antenna design and network conditions differ. Changes over time and correlation with latency are stronger evidence than one isolated value.
Interactive signal check
Enter the live SINR, RSRP and RSRQ values from the router dashboard. The score is a rough comparison aid for testing different router positions, not a guarantee of gaming quality or mast capacity.
When analytics is enabled, LinkSpeed records calculator use, validation errors, score band and verdict. Exact SINR, RSRP and RSRQ values are not sent to Google Analytics.
Estimated radio score: 0/100
The mobile link is not yet proven faulty. Return to the Wi‑Fi versus 5G comparison and improve the home wireless path.
Keep a short test log and compare another mobile network or fixed broadband option. Router placement may help only if it changes the serving band, sector or signal quality.
Retest several windows and heights. Consider the external antenna checks only after a repeatable position change improves the radio readings.
Loaded latency
Yes. 5G capacity can change from moment to moment, while uploads often have less headroom than downloads. When a router or mobile link queues too much traffic, gaming and calls can lag even though the connection still reports a high headline speed.
Pause backups, cloud sync, CCTV and livestreaming. Where the router supports it, apply SQM or QoS below the repeatable upload capacity rather than the highest one-off result.
Run the test more than once because the available 5G rate changes. Shape against a conservative repeatable speed and confirm that the router can manage queues at that throughput.
Move to the Packet Loss Test and compare game server regions. Queue control cannot fix radio loss or a poor destination route.
Gaming and NAT diagnosis
Treat latency and NAT as separate checks. Mast load, radio quality, packet loss and bufferbloat usually cause lag; CGNAT or local double NAT more often affects hosting, peer-to-peer sessions, matchmaking and party chat.
Many mobile networks share one public IPv4 address between multiple customers. The 5G router receives an address from a private or shared range, and the carrier performs another translation inside its core network.
Because the public address is not assigned directly to the home router, local port forwarding, UPnP or DMZ settings cannot create a true inbound path through the provider's upstream NAT.
Double NAT occurs when a separate gaming or mesh router is connected behind the 5G gateway while both devices still perform routing, DHCP and firewall functions.
This can complicate UPnP, inbound mappings, console NAT tests and peer-to-peer services. It does not automatically create large ping increases, but it can cause connection and matchmaking problems.
Common local private ranges include 10.0.0.0/8, 172.16.0.0/12 and 192.168.0.0/16. Carrier-grade NAT commonly uses 100.64.0.0/10.
A router WAN address inside one of these ranges is evidence of private addressing, but confirm it against the public address seen by an external IP-check service.
Where the provider, router, console and game support IPv6 correctly, devices can have globally routable IPv6 addresses without relying on IPv4 port translation. Firewall policy still applies, and not every game service uses IPv6 end to end.
Find the cellular or WAN IPv4 address. Menu names vary by router, so use the status, internet, mobile network or connection-information page.
Use the LinkSpeed IP & CGNAT checker. If the two addresses differ and the router WAN address is private or within 100.64.0.0/10, the service is likely behind CGNAT.
If the gaming or mesh router receives an address such as 192.168.x.x from the 5G gateway, the home has a local double NAT layer as well.
CGNAT usually affects inbound hosting, peer-to-peer sessions, NAT type and some voice or lobby services. It is not the normal cause of high download latency or radio jitter.
The connection is not behind IPv4 CGNAT at that moment. Investigate local double NAT, firewall rules, UPnP and the game service.
Port forwarding on the home router cannot bypass the upstream carrier translation. Ask the provider whether a public IPv4, business APN or suitable IPv6 service is available.
Remove the local double NAT by using bridge or modem mode where supported, or place the secondary router in access-point mode. A DMZ can simplify forwarding but does not remove carrier-grade NAT.
APN warning: do not copy an APN from an old forum post. Public-IP APNs, eligibility and charging vary by network, SIM type and contract. Use only the provider's current documented profile.
After changing APN, bridge mode, access-point mode or local router settings, rerun the console's own network test. Use the result to confirm the fault boundary rather than treating a NAT label as proof of radio lag.
NAT Type 1: the console is directly exposed without a normal router NAT layer. This is uncommon and is not required for good gaming.
NAT Type 2: the normal target for a console behind one correctly configured router.
NAT Type 3: restrictive translation or firewall behaviour may interfere with some peer-to-peer, party or hosting functions.
Open: the console has broad connectivity for multiplayer and hosting.
Moderate or Strict: some peer-to-peer, hosting or chat features may be limited, but the result does not by itself identify CGNAT.
Double NAT detected: check whether a second router is operating behind the 5G gateway. Use bridge mode where supported or place the secondary device in access-point mode.
The local topology was contributing to the console warning. Retest matchmaking, voice chat and hosting rather than relying only on the status label.
Compare the router WAN and public IPv4 addresses. A private or shared WAN address points towards CGNAT or another upstream translation layer.
Repeat the test over Ethernet and at another time of day. Packet loss can come from Wi-Fi, mast load, radio quality or provider routing and is separate from NAT type.
Do not chase NAT Type 1 on PlayStation. NAT Type 2 behind a correctly configured router is normally suitable. Exposing a console directly can remove useful firewall protection without improving radio latency or mast congestion.
Advanced checks
Use these only after the quick comparison identifies a matching symptom. They are not universal gaming tweaks.
Signal hardware
An external antenna can help when repeat tests show that weak or noisy radio conditions remain the limiting factor after careful router repositioning. Buy only after checking connector type, supported bands, MIMO layout, cable length and whether the router actually enables its external antenna ports.
A directional panel concentrates reception in one direction. It is most useful when the serving mast or sector is known and the property is near the edge of coverage or shielded by thick walls.
It requires careful alignment and can perform worse if pointed at the wrong sector or if the modem relies on multiple bands arriving from different directions.
An omnidirectional antenna receives from a wider area and is easier to install where the serving cell is uncertain or reflections are important.
It usually offers less focused gain and may collect more interference in dense urban areas.
A 4x4 antenna and router can use more independent radio paths than a 2x2 setup, but only when the router exposes four compatible ports and the network and band combination support it.
Do not assume a 4x4 antenna will improve a router that only accepts two external feeds.
Long, thin coaxial runs can lose a large amount of signal at 3.5 GHz and above. Keep cable runs short, use suitable low-loss cable and avoid unnecessary adapters.
| Antenna profile | Best fit | Main trade-off | What to validate |
|---|---|---|---|
| 4x4 directional MIMO panel | Known mast direction, weak edge-of-coverage signal, thick building fabric | Requires precise alignment and four compatible router ports | Repeatable improvement in SINR, RSRP, band stability and latency |
| 2x2 directional MIMO panel | Routers with two external ports and a known serving direction | May limit available spatial streams compared with an internal 4x4 system | Cleaner SINR and lower peak-time jitter without losing useful bands |
| 4x4 omnidirectional MIMO | Uncertain serving direction, urban reflections, multiple candidate sectors | Lower directional gain and greater exposure to local interference | Stable Cell ID, carrier aggregation and throughput across several positions |
Check the manual for connector type, number of ports, supported frequencies and any software setting that switches between internal and external antennas.
Use the router's Cell ID, band information and several tested positions. Public mast maps can help, but they may be incomplete or out of date.
Use suitable exterior hardware, weatherproof connections and professional installation where work at height or external cabling is involved.
Make small alignment changes, allow the modem time to settle, and record Cell ID, bands, RSRP, RSRQ, SINR, throughput and latency after each change.
A one-off speed increase is not enough. Prefer the position that improves signal quality and latency consistently across peak and off-peak tests.
Do not buy by claimed gain alone. Connector compatibility, cable loss, router MIMO support and the serving band's frequency matter more than a large marketing number.
Advanced router topology
A separate cellular gateway can handle the radio link while OpenWrt manages the home LAN, firewall and SQM. The cleanest arrangement is bridge mode or IP passthrough, but behaviour varies widely between modem vendors and mobile networks.
5G gateway: radio modem in bridge or IP-passthrough mode.
OpenWrt WAN: DHCP client on the physical port connected to the gateway.
OpenWrt LAN: a different private subnet, such as 192.168.1.0/24.
IP passthrough can remove the gateway's local NAT layer, but the OpenWrt WAN may still receive a private or 100.64.0.0/10 address from the mobile provider.
That means local double NAT is gone while upstream carrier-grade NAT remains.
Some bridged modems keep a management address such as 192.168.8.1. Others require a vendor-specific route, separate VLAN or do not expose the dashboard in bridge mode at all.
Do not use the same subnet on both sides. For example, keep the modem on 192.168.8.0/24 and the OpenWrt LAN on 192.168.1.0/24.
Export a backup before changing interface or firewall settings so the router can be restored if management access is lost.
Map it to the physical Ethernet port connected to the 5G gateway and place it in the standard WAN firewall zone.
Check whether OpenWrt receives a public address, a private address or a carrier-grade NAT address.
If the modem supports a reachable management subnet, add a host or subnet route according to that device's documentation rather than copying a generic secondary-interface recipe.
Confirm internet access, DNS, modem dashboard access and the absence of local double NAT. Then apply SQM to the real bottleneck interface and use the path MTU and MSS checks if TCP sessions stall or fragment.
ip addr show
ip route show
ubus call network.interface.wan status
nft list ruleset
Use the output to confirm the WAN address, default route, modem-management route and active firewall rules.
Use OpenWrt's normal WAN zone defaults: unsolicited input and forwarding should be rejected or dropped, outbound traffic allowed, masquerading enabled where NAT is required, and only documented exceptions added.
Do not append a second full config zone block with the same name. Edit the existing WAN zone so that network assignments and defaults are not duplicated.
LAN clients usually need outbound access to the modem dashboard, not a rule allowing every packet from the modem subnet into the LAN. Stateful firewall return traffic already handles replies to connections initiated from LAN.
Allowing selected ICMP types can aid path MTU discovery and diagnostics. Matchmaking does not generally require unrestricted inbound ping, so do not add a rule solely on that assumption.
On firewall4 systems, inspect nft list ruleset and the LuCI firewall status. Confirm only the intended rules are matching and that the WAN input policy remains restrictive.
Version warning: OpenWrt firewall syntax changed from iptables-based firewall3 to nftables-based firewall4. Follow the documentation for the installed release and avoid copying legacy chain names.
Packet sizing
Cellular links can use extra encapsulation inside the provider network. When the usable path MTU is lower than expected, oversized packets may fragment or fail, causing slow transfers, stalled connections or erratic application behaviour. Do not assume every 5G service uses the same MTU.
MTU is the largest IP packet size that can cross a link without fragmentation. MSS is the maximum TCP payload advertised during the handshake.
MSS clamping helps TCP sessions avoid oversized packets. It does not rewrite UDP game packets, so it cannot directly fix every gaming issue.
IPv4 and IPv6 rely on control messages to report that a packet is too large. Blocking required ICMP messages can create a path MTU black hole where some websites, VPNs or game services partially stall.
The correct MTU depends on the modem, provider, access mode, IPv4 or IPv6 path and any PPPoE, VPN or tunnel overhead. Bridge mode alone does not prove that 1420 or 1460 is correct.
Some hosts rate-limit or ignore diagnostic pings. Confirm the result against at least two stable destinations before changing the router.
Windows: ping 1.1.1.1 -f -l 1472
macOS: ping -D -s 1472 1.1.1.1
Linux: ping -M do -s 1472 1.1.1.1
Reduce the payload until the test succeeds, then add 28 bytes for an IPv4 ICMP estimate. For IPv6, header sizes and commands differ, so do not reuse the same equation blindly.
This corresponds to a 1500-byte IPv4 packet after adding the 20-byte IP header and 8-byte ICMP header.
Drop by 10 or 20 bytes until replies succeed, then increase in 1-byte steps to find the highest repeatable payload.
A single silent host is not proof of a lower MTU.
Use MTU or MSS adjustments for repeatable fragmentation, stalled TCP sessions or proven path MTU problems, not as a generic gaming tweak.
On OpenWrt, the normal WAN-zone option is mtu_fix '1'. In LuCI this is usually shown as MSS clamping. It generates TCP MSS adjustment rules based on the interface path and is preferable to inventing a static value first.
uci show firewall | grep -E 'wan|mtu_fix'
nft list ruleset | grep -i -E 'mss|mtu'
ip link show
After enabling MSS clamping, restart the firewall with /etc/init.d/firewall restart and repeat the affected TCP test.
Avoid undocumented firewall options. A setting such as mss_clamping_fixed is not a standard portable OpenWrt UCI option across releases. Use the supported mtu_fix behaviour or an explicit nftables rule only when it has been tested for the installed firewall version.
Cell change diagnostics
A sudden cell, sector or band change can coincide with a lag spike, but modem APIs are vendor-specific and often require authentication. Use a maintained integration for the exact modem model rather than assuming that a generic URL such as /api/device/signal exists.
Useful fields include Cell ID, PCI, NR-ARFCN or EARFCN, active LTE and NR bands, RSRP, RSRQ, SINR and the timestamp of the change.
Normal mobility, carrier aggregation and network optimisation can change serving cells or bands without harming performance. Correlate the event with packet loss, latency and the game timestamp.
Many gateways have slow web interfaces. A five-second unauthenticated polling loop can create extra CPU load, lockouts or log noise. Start with 30–60 second intervals unless the vendor documents a safe API rate.
OpenWrt services should normally run under procd so they can restart cleanly, stop during shutdown and write predictable logs.
Use vendor documentation, a supported OpenWrt package, AT commands, ModemManager, uqmi, umbim or a documented local API for that exact hardware.
Confirm the output format and identify stable field names before writing any parser.
Store the previous Cell ID, PCI and band, then write to syslog only when one of those values changes.
Compare the log timestamp with gateway ping, external ping and the game's own network telemetry.
logread -f | grep -i -E 'modem|wwan|cell|handover|5g'
ubus list | grep -i -E 'modem|network'
ubus call network.interface.wan status
If the router uses ModemManager, inspect the modem with mmcli -L and the appropriate mmcli -m command for the detected modem index.
Do not deploy a generic scraper as a production daemon. Hard-coded endpoints, brittle grep/sed parsing and background launch from rc.local can silently fail or misreport Cell IDs. Build the logger around the actual modem interface and supervise it with procd.
FAQ
Speed and latency measure different things. A fast 5G download can still have unstable ping, jitter or packet loss because of mast load, changing radio quality, Wi‑Fi or queues filling under load.
Repeated evening spikes can indicate heavier mast or provider-network load. Compare the same Ethernet ping test off peak and during the normal problem period before changing hardware.
Compare the same device over Ethernet and Wi‑Fi. Stable Ethernet with poor Wi‑Fi points to the home wireless path; instability on both keeps the 5G radio and provider path in scope.
Yes. Upload or download queues can increase ping when the link is busy, especially because available 5G capacity changes. A bufferbloat test shows whether loaded latency rises sharply.
It can be suitable when Ethernet ping, jitter and packet loss remain stable at the times you play. Full fibre is usually more predictable where competitive latency matters.
CGNAT mainly affects inbound hosting, NAT type and some peer-to-peer services. It is not normally the main cause of radio jitter or high loaded latency, so test ping and packet loss separately.