LoRaWAN Antenna Theory and What Actually Matters

Most LoRa guides spend a lot of time on gateways, network servers, and application integrations. The antenna gets a paragraph at most, usually something like “attach the included antenna before powering on.” That advice is technically correct but leaves out the part that actually determines how well your network works.

The antenna is where the radio signal meets the physical world. Everything upstream of it, the transmit power, the spreading factor, the sensitivity of the receiver, is fixed by the hardware and the protocol. The antenna is the one variable in the link that you directly control, and making a poor choice there cannot be compensated for anywhere else in the system.

This article covers what you actually need to understand about antennas to make sensible decisions for a LoRaWAN deployment. It is not a complete RF engineering course. It is the subset of antenna theory that has practical consequences for someone setting up a gateway and a handful of sensors.

Gain, and what it costs

Antenna gain is the single most misunderstood number in LoRa hardware marketing. It sounds straightforwardly good: more gain means better range, right? Mostly, but with an important catch.

Antennas do not amplify. They cannot. An antenna is a passive device with no power source of its own. What gain actually describes is concentration: a high-gain antenna takes the energy available to it and focuses it into a narrower beam, producing a stronger signal in that direction at the cost of reduced signal in other directions.

The reference point for gain is an isotropic radiator, a theoretical antenna that radiates equally in all directions like a perfect sphere. This does not exist in practice, but it is a useful baseline. Gain measured relative to it is expressed in dBi. A 0 dBi antenna radiates equally in all directions. A 3 dBi antenna doubles the effective power in its favoured direction by concentrating energy that would otherwise have gone elsewhere.

For a vertical omnidirectional antenna, which is what almost every LoRa gateway and sensor uses, the concentration happens in the horizontal plane. The antenna radiates a disc of energy rather than a sphere. As gain increases, the disc gets flatter and wider: a 6 dBi antenna covers more horizontal distance than a 3 dBi antenna, but its signal drops off faster above and below. A 9 dBi antenna covers even more horizontal distance, but its disc is so flat that nodes on a hill above the gateway or in a basement below may be partially or entirely outside the coverage area.

The practical implication: for a gateway mounted on a rooftop or high elevation point serving sensors spread across flat terrain, higher gain is straightforwardly better. For a gateway inside a building serving sensors on multiple floors or at varying elevations, a 3 dBi antenna often outperforms a 9 dBi one because the lower-gain antenna’s signal reaches above and below where the high-gain antenna’s disc does not.

dBi versus dBd

You will occasionally see antenna gain expressed in dBd rather than dBi. The reference point is different: dBd measures gain relative to a half-wave dipole antenna rather than a theoretical isotropic radiator. A half-wave dipole has 2.15 dBi of gain by definition, so converting between the two is straightforward: dBi = dBd + 2.15.

A 3 dBd antenna is the same as a 5.15 dBi antenna. Marketing materials sometimes use dBd when the dBi figure would look less impressive. The number to use when comparing antennas is always dBi; if a spec sheet only gives dBd, add 2.15 to get the equivalent dBi.

VSWR: the number that tells you if the antenna actually works

Voltage Standing Wave Ratio is less well-known than gain but arguably more important for practical deployments. It measures how efficiently the antenna transfers power from the radio to the air at the target frequency.

When an antenna is not well matched to its operating frequency, some of the transmit power reflects back down the cable toward the radio rather than being radiated. This is called a standing wave, and the ratio of forward power to reflected power is the VSWR. A perfect match is 1:1. In practice, anything below 1.5:1 is excellent. Below 2:1 is acceptable. Above 2:1, you are losing a meaningful fraction of your transmit power to reflection, and above 3:1 you may be damaging the radio over time.

The relevance for LoRa: many inexpensive antennas sold as “868 MHz” are actually wideband designs covering a broad range of frequencies. They are tuned to be acceptable across 860 to 930 MHz rather than optimised for exactly 868 MHz. The VSWR at precisely 868 MHz on these antennas can be 2.5 or higher, meaning a significant portion of your transmit power never leaves the device.

This is not theoretical. Community antenna testing projects for Meshtastic and LoRa have measured budget antennas with VSWR above 8 at 868 MHz, where the majority of transmit power reflects straight back into the radio. The antenna looks fine; the radio works; the range is inexplicably poor. VSWR is why.

When evaluating an antenna, look for a VSWR specification at 868 MHz specifically, not across a broad frequency range. Anything below 1.5:1 at 868 MHz is a well-tuned antenna. Anything without a frequency-specific VSWR figure is a marketing document, not a datasheet.

Cable loss and why it matters more than you expect

Coaxial cable between the radio and the antenna attenuates the signal in both directions. The loss is frequency-dependent: higher frequencies lose more signal per metre of cable than lower ones.

At 868 MHz, common coaxial cable types lose approximately:

• RG-174 (thin, flexible, commonly used in short pigtails): around 1.5 dB per metre
• RG-58 (medium, commonly used in amateur radio): around 1.2 dB per metre
• LMR-200 (better quality, stiffer): around 0.5 dB per metre
• LMR-400 (high quality, thick): around 0.22 dB per metre

These numbers matter because a LoRa link budget is tight. If you mount a gateway antenna on a mast on the roof and run 5 metres of RG-58 back to the gateway radio, you have lost 6 dB before the signal reaches the antenna. That is the equivalent of reducing your transmit power to one quarter, or taking a theoretical 3 dBi antenna and turning it into a -3 dBi one.

The general principle: minimise cable length, and use the best quality cable you can justify for the run length involved. For a gateway where the radio and antenna are close together, the included pigtail is fine. For any installation where the antenna needs to be meaningfully separated from the radio, budget for good cable and factor the loss into your coverage planning.

The link budget

The link budget is the accounting exercise that tells you whether two devices can hear each other. It adds up all the gains and subtracts all the losses in the path between transmitter and receiver.

A simplified LoRaWAN link budget looks like this:

EIRP = Transmit power (dBm) + Antenna gain (dBi) - Cable loss (dB)

Received signal = EIRP - Path loss (dB) + Receive antenna gain (dBi)

Link margin = Received signal - Receiver sensitivity (dBm)

In the UK, the maximum permitted transmit power at 868 MHz is 14 dBm (25 mW) for duty cycles above 1%, and 16 dBm for shorter bursts. A LoRa radio at maximum legal power with a 3 dBi antenna and 1 dB of cable loss produces an EIRP of 16 dBm.

LoRa receiver sensitivity at common spreading factors is around -120 to -137 dBm depending on the spreading factor and bandwidth. A higher spreading factor gives better sensitivity but slower data rate.

Path loss over distance is significant. At 868 MHz in free space, the signal loses roughly 92 dB over 1 km. Real-world conditions with buildings, trees, and terrain add further loss on top of that.

The upshot: LoRaWAN links have a link budget of around 155 dB at high spreading factors, which explains the multi-kilometre range figures you see in datasheets. In practice, urban environments with heavy obstruction bring that down considerably.

What this means for a practical deployment

For sensors: the included antenna is usually a quarter-wave stubby with 1 to 2 dBi of gain and questionable VSWR. For a sensor that will live near the gateway or in a location with good line of sight, it is adequate. For a sensor that will be in a challenging location, a better antenna matters. Check whether the sensor uses an SMA connector or a proprietary one before assuming you can swap it out.

For gateways: the antenna choice has more impact than any other hardware decision you will make. A gateway with a 3 dBi well-tuned antenna in a high location will outperform a gateway with a 9 dBi poorly-tuned antenna at ground level. Height and line of sight matter more than the gain figure on the box.

For cable runs: if the gateway radio needs to be separated from the antenna, budget for LMR-200 or better for any run over a metre. The cost difference between mediocre and good cable is modest; the performance difference is not.

For the sensor that refuses to connect: before adjusting spreading factors or troubleshooting the network stack, check the antenna. A disconnected antenna, a cracked connector, or a badly-tuned stubby is frequently the cause of range problems that appear to be software issues.

The one thing to do first

Before anything else in a LoRa deployment, put the gateway antenna somewhere high with a clear view of the sky and as much unobstructed horizontal distance as possible. No amount of careful network configuration compensates for an antenna that cannot see the sensors it is supposed to serve.