Puncturing in WLAN PHY – What It Really Means

Let’s get this out of the way first: “puncturing” is one of those terms that sounds like you're breaking something. And in a way – you are. You’re chopping out parts of a signal (either in time or frequency) to make the whole thing fit better – either for performance, compatibility, or because the real world just isn’t clean and perfect.

1. Coding-Time Puncturing – Dropping Bits on Purpose

In older and current OFDM PHYs (think 802.11a/n/ac/ax), data is first encoded with a convolutional encoder, typically at a base rate like R = 1/2. Now, if you want to increase the data rate (but keep some error protection), you don’t re-encode – you just skip some of the encoded bits. That’s puncturing. You remove bits according to a pre-defined pattern. Receiver knows that, fills in the blanks (kind of) with soft-decision decoding.

Why? Because we want higher throughput but don’t want to rebuild the whole modulation scheme. Smart hack.

2. Channel Puncturing – Cutting Frequency Holes

Welcome to HE (802.11ax) and EHT (802.11be). We’re working with big channel widths – 80, 160, 320 MHz. But sometimes, parts of that spectrum are full of garbage (interference, DFS, legacy neighbors, or region restrictions). What do we do? We puncture the bad subchannels. Just don’t use them.

Granularity:

• 20 MHz chunks are the base unit for everything up to 160 MHz
• 40 MHz chunks when working with 320 MHz

This isn't just "don’t transmit there". The PPDU actually tells the receiver what's being used and what's not – encoded in fields like U-SIG, Bandwidth Indication, and Punctured Channel Information.

3. Preamble Puncturing – Yes, Even the Header Gets Poked

Sometimes it’s not just the data payload – we need to puncture the preamble too. This makes things trickier, because that’s what tells everyone “Hey! I’m transmitting!”

Special modes like HE-CBW-PUNC80-SEC and HE-CBW-PUNC160-PRI20 describe exactly which 20 MHz pieces get cut. And the spectral mask – the shape of your transmission in frequency domain – needs to reflect this. No leaking into forbidden zones.

If two masks (main + puncture) define limits, take the stricter one.

4. Punctured Sounding – Only Beamform What Matters

In beamforming (Sounding, NDP), feedback isn’t useful for parts of the channel we’re not even using. So we tell the receiver “Hey, don’t bother sending CSI for these punctured parts.”

How?

• HE: Use the Disallowed Subchannel Bitmap in the HE NDP Announcement.
• Beamformee echoes that in the feedback.
• EHT: Even tighter. If a subchannel is punctured in U-SIG, it’s off-limits. No feedback requested or allowed.
• 242-tone RUs overlapping with punctured 20 MHz chunks? Also ignored.

5. Impact on Channel Access – CCA & RTS/CTS Game

When you receive an RTS on a channel that might include punctured bits, you can’t just respond. You need to make sure that all nonpunctured 20 MHz subchannels were clear (CCA idle) for a full PIFS duration before answering with CTS.

Why? Because someone else might be using one of those side channels, and they didn’t hear your RTS if it was punctured. Play fair.

Remember

• PHY coding puncturing: drop coded bits to get better rates, decoder fills the gaps.
• Frequency puncturing: skip bad subchannels (interference, DFS, etc.) while keeping wide channels functional.
• Preamble puncturing: header gets trimmed too – spectral masks must comply.
• Sounding with puncturing: beamforming feedback ignores dead zones.
• Channel access: CCA respects only the active subchannels, not the ones you ignore.

Puncturing is clever engineering. Don’t waste spectrum. Work around noise. Respect legacy. Stay fast.