RF Link Budget Analysis: Engineering Guide

What Is a Link Budget?

Think of a link budget as a careful accounting ledger for your RF signal — every gain and loss from the transmitter output all the way to the receiver input. The question you're answering is simple: does enough power arrive at the receiver for it to decode the signal? If your received power beats the receiver sensitivity by a comfortable margin, you're golden. If not, you've got work to do — add more transmit power, swap in better antennas, trim cable losses, or just move the radios closer together.

Most engineers I've worked with treat link budgets like an afterthought until a prototype fails in the field. Don't be that person. A solid link budget catches problems on paper before you've committed to hardware.

The Fundamental Equation

The whole analysis boils down to one equation. Everything's in dBm or dB, which makes the math wonderfully simple — just addition and subtraction:

The result is your received power in dBm. You start with transmit power, add antenna gains, subtract all the losses along the way, and what's left is what arrives at the receiver.

Link margin is just the difference between what you receive and what you need:

Link margin = P_rx − Sensitivity_rx

Positive margin means the link works. But how much margin do you actually need? That depends on your application and how much you trust your environment:

• Indoor WiFi typically needs 10–15 dB margin. Plenty of multipath, people moving around, interference from neighboring networks.
• Outdoor point-to-point links usually want 15–20 dB. Weather changes propagation, rain attenuates the signal, and you need headroom for the occasional tree growing into your Fresnel zone.
• Satellite links often run with just 3–6 dB margin because every additional dB costs real money in transmitter power, antenna size, or both. When you're launching hardware into orbit, you optimize ruthlessly.

Free-Space Path Loss

FSPL dominates every single wireless link budget. It's the biggest number you'll deal with, and it grows fast with distance and frequency. Here's the thing though — it's not really a "loss" in the sense that something's absorbing your signal. It's pure geometry. Your transmitter radiates power in all directions (or at least some solid angle), and the power density drops as the wavefront expands. By the time it reaches the receiver, you're collecting only a tiny fraction of what was transmitted.

The equation looks like this:

Where d is distance, f is frequency, and c is the speed of light. For quick mental math when you're sketching out a link on a whiteboard, this approximation is close enough:

FSPL ≈ 20log(f_GHz) + 20log(d_km) + 92.4 dB

Let me give you some concrete numbers so you can build intuition:

• 2.4 GHz at 100 meters: 80 dB of path loss
• 2.4 GHz at 1 kilometer: 100 dB of path loss
• 28 GHz (5G mmWave) at 100 meters: 101 dB — that's 21 dB more loss than 2.4 GHz at the same distance

That last one explains why 5G mmWave coverage is so challenging. The path loss scales with the square of frequency, so when you jump from 2.4 GHz to 28 GHz, you're fighting an uphill battle. The only way to compensate is with high-gain antennas and beamforming, which is exactly what 5G systems do.

Receiver Sensitivity

Your receiver sensitivity sets the floor for how weak a signal you can successfully decode. It's determined by two things: the noise floor of your receiver and the signal-to-noise ratio your modulation scheme needs to hit an acceptable bit error rate.

The equation is:

Let's break down each term:

• −174 dBm/Hz is the thermal noise power spectral density at room temperature. It comes from kT, where k is Boltzmann's constant and T is 290 K. This is physics — you can't get around it without cooling your receiver.
• BW is your receiver bandwidth in Hz. Wider bandwidth means more noise gets in. This is why narrowband systems like LoRa can achieve incredible sensitivity.
• NF is the noise figure of your receiver in dB. A perfect receiver would have 0 dB noise figure, but real receivers add noise. Consumer WiFi chipsets are typically 5–8 dB. High-end spectrum analyzers might hit 3 dB. Low-noise amplifiers for satellite ground stations can get below 1 dB, but they cost thousands of dollars.
• SNR_min is the minimum signal-to-noise ratio your demodulator needs. Simple modulation like BPSK might only need 10 dB. Dense modulation like 64-QAM needs 25 dB or more. There's always a tradeoff between data rate and sensitivity.

Here's a real example: a typical 802.11n receiver operating in a 20 MHz channel with a 7 dB noise figure and requiring 10 dB SNR for the lowest data rate:

S_min = −174 + 10log(20×10^6) + 7 + 10 = −174 + 73 + 7 + 10 = −84 dBm

That −84 dBm is what the WiFi spec calls out as the minimum sensitivity for the lowest MCS index. Higher data rates need better SNR, so sensitivity gets worse (less negative) as you step up through the modulation and coding schemes.

Worked Example: 900 MHz IoT Link

Let's work through a complete link budget for a realistic system. Say you're designing a 900 MHz IoT sensor network with a 500 meter range requirement. You're using a LoRa-style modulation for its excellent sensitivity. The environment is outdoor with some light foliage and buildings in the path.

Walk through the math: start with +20 dBm transmit power, add 2 dBi antenna gain, subtract 0.5 dB cable loss on the transmit side. That gives you +21.5 dBm EIRP leaving the transmit antenna. Then subtract 85.7 dB for free-space path loss and another 5 dB for environmental effects. You're down to −69.2 dBm arriving at the receive antenna. Add 2 dBi receive antenna gain and subtract 0.5 dB cable loss, and you end up with −67.7 dBm at the receiver input.

The LoRa receiver at spreading factor 7 has a sensitivity around −123 dBm. Your margin is a whopping 55.3 dB. That's honestly overkill for most applications. You could extend the range to several kilometers, or you could back off the transmit power significantly to save battery life. At 0 dBm transmit power (1 milliwatt), you'd still have 35 dB of margin, which is plenty for a reliable link with some fade margin built in.

Common Mistakes

After reviewing dozens of link budgets from other engineers, I've seen the same mistakes over and over. Here are the ones that hurt the most:

Forgetting polarization loss. Antenna polarization matters more than most people think. If your transmit antenna is vertically polarized and your receive antenna is horizontally polarized (90° cross-polarization), you lose around 20 dB. Even partial misalignment costs you. Two linear antennas at 45° relative rotation lose about 3 dB. This is especially common in mobile applications where the receiver orientation isn't controlled. Sometimes circular polarization is worth the 3 dB penalty compared to linear just to avoid this problem entirely. Ignoring impedance mismatch. Every connector, every cable, every transition in your RF chain needs to be impedance matched. A 2:1 VSWR creates about 0.5 dB of mismatch loss. That might not sound like much, but in a tight link budget where you're fighting for every dB, half a dB matters. I've seen systems fail in the field because someone used a cheap cable with poor return loss. The individual reflections were small, but they added up across multiple connectors. Using peak antenna gain in all directions. This one catches a lot of people. Antenna gain is directional. When the datasheet says your patch antenna has 6 dBi gain, that's only true in the boresight direction — straight ahead. Move 30° off-axis and you might be down to 0 dBi. Move 90° to the side and you could be at −10 dBi or worse. If your link geometry isn't perfectly aligned (and in the real world, it rarely is), you need to account for the actual gain in the direction of the link, not the peak gain from the datasheet. Not accounting for fading margin. This is the one that bites you in production. Your link budget might look perfect in free space, but real wireless channels fade. Multipath propagation creates deep nulls where signals cancel. Moving objects cause Doppler shifts and time-varying fading. For indoor or urban environments with rich multipath, you should add 10–15 dB of fade margin on top of your basic link margin. For line-of-sight outdoor links, 5–8 dB is usually enough. Satellite links with clear sky conditions might only need 3 dB. The point is, don't design for the average case — design for the 99th percentile case where the channel is in a deep fade.

Use our RF Link Budget Calculator to model your system. It'll compute received power versus distance, show you where your margin runs out, and help you visualize how different parameters affect the link. It's a lot faster than doing the math by hand every time you want to try a different antenna or frequency.