Teltonika Signal Optimisation: How to Improve 4G and 5G Performance

The first mistake: treating signal strength as performance

When a 4G or 5G connection is slow, the obvious reaction is to look at the bars. Four or five bars feels good; one or two feels bad. That is useful as a quick indication, but it is nowhere near enough to diagnose a modern cellular link.

A cellular connection has both strength and quality. A router can receive a very powerful signal from a base station and still have poor throughput because that signal is noisy, interfered with or arriving from a heavily loaded cell. Equally, a weaker signal with a clean radio path can perform very well.

On a Teltonika router the useful measurements are normally available under the mobile status pages in RutOS. For LTE and 5G, the three figures to become familiar with are RSRP, RSRQ and SINR.

RSRP, RSRQ and SINR: what the numbers are telling you

Teltonika’s own signal guidance treats these measurements together rather than using RSSI alone. That distinction matters because RSSI includes the total received radio power, including interference. A big number can therefore look reassuring while disguising an ugly radio environment.

Those numbers are guidance, not pass/fail limits. A telemetry device sending a few kilobytes every hour can work reliably under radio conditions that would make a broadband user miserable.

Test the router, not your phone

This sounds obvious once it is pointed out, but it is one of the most common mistakes in cellular troubleshooting. Someone stands next to the router, connects a phone to its Wi-Fi, runs a consumer speed-test app and concludes that the 4G or 5G connection is slow.

That test contains two wireless links:

Mobile network → Teltonika cellular modem → Teltonika Wi-Fi → phone.

A disappointing result could therefore come from the mobile network, Wi-Fi interference, a busy Wi-Fi channel, distance from the access point, the phone’s own radio capability, power-saving behaviour or local traffic. You have introduced another variable before you have even measured the WAN.

Where supported, use the built-in RutOS Speed Test. Teltonika provides a Speed Test page on the RUTX50 specifically to measure WAN download and upload performance, select a test server and repeat the test without involving the client’s Wi-Fi connection.

One warning: a speed test moves a meaningful amount of data. On a metered IoT SIM, do not repeatedly run tests without checking the tariff and allowance first.

Your Teltonika is not simply connected to “the nearest mast”

The mental picture most people carry is simple: router, nearest tower, Internet. Cellular networks do not work quite like that.

The modem camps on a particular cell and sector using a particular radio layer, but that choice is dynamic. The serving network can change as signal conditions, network parameters, congestion, mobility and available frequencies change. With LTE Advanced and 5G, carrier aggregation and NSA operation can make the picture more complicated again because more than one carrier may be contributing to the connection.

A router that has not moved an inch can therefore behave differently tomorrow. A router on a vehicle or boat adds another layer because its physical radio environment is changing continuously.

QuWireless builds all-in-one outdoor antenna enclosures specifically for devices including the RUTX50. The router is installed within the antenna housing, reducing RF cable length and putting the cellular equipment at the useful outdoor antenna position. QuWireless is now part of Taoglas, following Taoglas’ acquisition of the Polish antenna-enclosure specialist in August 2026. See our Taoglas and QuWireless coverage and the QuWireless RUTX50 outdoor antenna/enclosure.

Why Band 20 can be brilliant today and Band 1 terrible tomorrow

Imagine checking the installation on Monday. The RUTX50 is on LTE Band 20 and the built-in test reports 70 Mbps. On Tuesday you return to apparently the same place. Signal bars still look respectable, but the router is now showing Band 1 and the speed test manages 8 Mbps.

It is tempting to conclude that Band 20 is “the fast band” and Band 1 is “the slow band”. That would be the wrong lesson.

Band numbers describe spectrum, not guaranteed performance. In the UK, Band 20 at 800 MHz has useful propagation and building penetration. Band 1 at 2100 MHz behaves differently. But throughput also depends on how much spectrum the operator has configured, the bandwidth of the particular carrier, how loaded the cell is, interference, radio quality, backhaul, carrier aggregation and which physical site or sector is serving you.

Tomorrow the opposite result could be true.

Why routers change bands and cells

A cellular modem is not running a permanent marriage with one tower. It is continually operating inside selection and reselection rules set by the network and implemented by the modem. Depending on technology and state, the network can influence which cells and layers are preferred, while the modem measures its radio environment and moves when the criteria tell it to.

That can happen because:

• the device physically moves;
• signal levels from neighbouring cells change;
• interference changes;
• the serving cell becomes loaded;
• the operator changes network parameters;
• a different frequency layer becomes preferable;
• 5G becomes available or unavailable;
• the modem re-registers after a restart or loss of service;
• carrier aggregation combinations change.

This also explains why rebooting a router can appear to “fix” performance. The reboot may simply have forced the modem to re-register and land on a different cell or radio combination. Unless you record the before-and-after network information, you do not know whether you fixed anything or merely rolled the cellular dice again.

Which tower am I actually using?

Once performance starts moving around, record the serving-network information from RutOS. Depending on modem and technology, useful fields can include operator/PLMN, Cell ID, TAC, PCI, LTE band, 5G band and radio measurements.

Then use that information to investigate the surrounding network. CellMapper is a particularly useful tool because it maps cellular sites, cells, sectors and observed coverage from crowdsourced measurements.

It can help answer questions such as:

• Is the router repeatedly moving between two sites?
• Does the cell ID correspond with the tower you assumed it was using?
• Is the good-performing cell in a different direction from the poor one?
• Is a lower-frequency layer arriving from much further away?
• Would moving or reorienting the antenna favour a cleaner serving sector?

CellMapper is not the operator’s engineering database. It is crowdsourced and coverage varies by area, so treat it as an investigative aid rather than absolute truth. But when the Cell ID changes at exactly the same time performance collapses, it gives you a very useful lead.

Build a log: the pattern often becomes obvious

For intermittent problems, a single snapshot is rarely enough. Record a small set of values whenever the connection is good and whenever it is bad.

Illustrative values only. The important point is the method. The apparently strongest RSRP in this example produces the worst download result because the quality and serving-cell conditions are different.

Do this for long enough and “random” behaviour often stops looking random.

Move the router or antenna before buying more gain

Before replacing anything, change the physical environment. For an indoor router, temporarily test near a window, upstairs, on another side of the building or away from metal cabinets and machinery. For an outdoor antenna, test orientation and position where practical.

Record the same measurements each time. You are looking for a cleaner radio path, not merely a bigger RSRP number.

This is one reason an integrated outdoor enclosure such as the QuWireless approach can work well. Long RF cable runs cost signal, especially as frequency increases. Putting the router physically close to the antenna and extending Ethernet or PoE back into the building avoids throwing away useful RF performance in coaxial cable.

Check antennas, MIMO and connectors

Modern LTE and 5G routers use multiple antenna paths for MIMO. The extra connectors are not decorative backups. Depending on the network mode, multiple radio paths contribute to throughput and robustness.

Check that all required cellular antenna connections are present, connectors are secure, cables are undamaged and the antenna arrangement matches the manufacturer’s design. If the router is inside a metal enclosure, the antennas need to be outside it or specifically engineered as part of the enclosure.

For directional antennas, do not blindly point at “the nearest mast”. Establish which cell and sector you actually want before optimising orientation. The nearest tower may not be your operator’s best serving site, and the strongest cell may not be the least congested one.

Scan networks if the SIM gives you a choice

With a roaming or multi-network IoT SIM, an operator scan can be one of the most revealing tests available. A device may be able to see several networks at the same location, and the one it automatically selects is not guaranteed to produce the highest throughput for your application.

You might find one operator at -76 dBm RSRP with poor SINR and another at -86 dBm with much cleaner quality. The second network can easily be the better connection.

Manual operator selection is useful for testing, but be cautious about turning a diagnostic test into a permanent lock. Locking a multi-network SIM to one PLMN can remove the resilience you bought the SIM for in the first place.

Band locking: useful test, dangerous reflex

Band locking is valuable when used deliberately. If performance repeatedly collapses whenever the router lands on a particular LTE layer, temporarily testing individual bands can help isolate the behaviour.

But do not jump from “Band 3 tested well today” to “lock every router permanently to Band 3”. Networks evolve. Spectrum is refarmed, cells are reconfigured and operators change carrier combinations. A lock that fixes a deployment today can be the reason it stops working properly in the future.

Use a band lock to answer a question. Only keep it when you have a clear operational reason and a plan to revisit it.

Strong signal, good SINR, still slow? Stop blaming the antenna

If the router shows healthy RSRP, RSRQ and SINR but the built-in WAN speed test is still poor, the radio link may not be the problem.

Consider:

• cell congestion;
• operator traffic management;
• the SIM tariff or APN;
• VPN overhead;
• the chosen speed-test server;
• backhaul limitations;
• local routing or firewall configuration;
• the application server itself.

Also test at different times of day. If the same cell, band and similar signal measurements deliver 70 Mbps in the morning and 7 Mbps during the evening peak, that is a very different diagnosis from an antenna that suddenly loses 15 dB of RSRP.

The engineer’s rule: change one thing

The easiest way to make cellular troubleshooting impossible is to change the antenna, move the router, lock the band, change the APN and reboot the modem all at once.

If the connection improves, you still have no idea why.

A better method is almost deliberately boring: measure, change one thing, measure again.

The Teltonika already gives you much of the instrumentation you need. Add serving-cell information, a tower map and a disciplined test method and a frustrating “sometimes it is fast, sometimes it is slow” problem becomes an engineering investigation rather than guesswork.

Frequently asked questions