# Why do the OFDM training (and payload) symbols have silent sub-carriers (aka virtual carriers)?

I am examining an OFDM signal.

Training symbol:

• The first two OFDM symbols are training symbols, they are 2PSK modulated (within the subcarriers).
• In the first OFDM symbol the $(0+18N)^{\rm th}$ subcarriers are silent ($N$ is integer).
• In the second OFDM symbol the $(9+18N)^{\rm th}$ subcarriers are silent.

• In the first payload symbol, i.e. in the $3^{\rm rd}$ OFDM symbol, the $(6 + 18N)^{\rm th}$ subcarriers are silent.
• In the second payload symbol, i.e. in the $4^{\rm th}$ OFDM symbol, the $(6 +9 + 18N)^{\rm th}$ subcarriers are silent.
• In the third payload symbol, i.e. in the $5^{\rm th}$ OFDM symbol, the $(3 + 18N)^{\rm th}$ subcarriers are silent.
• In the $4^{\rm th}$ payload symbol, i.e. in the $6^{\rm th}$ OFDM symbol, the $(3 +9 + 18N)^{\rm th}$ subcarriers are silent.
• The payload symbols are QPSK modulated.

Then this pattern seems to repeat.

Pattern:

• So to summarise, the offsets are : $0, 9, 6, 6 + 9, 3, 3 + 9, 0, 9, \ldots$

Questions:

• How does this make sense ?
• What purpose does such a unique silencing of subcarriers serve ?
• I suspect it has to do with equalizations , but why is silencing useful ?
• And why this odd way of silencing the subcarriers ?
• Is this (way of silencing subcarriers) somehow connected to the Schmidl and Cox method ?

EDIT: Google gives this on "OFDM silent subcarriers"

EDIT2: The signal has been emitted by a MIMO system.

EDIT3: This is the Google search for "OFDM virtual carriers".

EDIT4: This is a 1024 carrier OFDM signal.

• Is there a constant pattern of how many subcarriers are unused in the payload data? Jun 1 '16 at 14:58
• @MarcusMüller Yes, I've edited the question to include the payload symbols too. It seems that every 18th subcarrier is silent. Jun 2 '16 at 7:24
• @jhegedus: Note that the typical term for unused carriers in OFDM is virtual carriers. A Google search yields many mentions in the literature on methods for exploiting virtual carriers to aid in synchronization. That could be one justification for including them. Jun 2 '16 at 13:04

• If you have to deal with OFDM it is best to have some background in signal theory/communication engineering. Basically you have to look at synchronization and equalization techniques. The spread of the energy loss over a large spectrum is an educated guess but could work this way. Also I guess you could develop some kind of tracking of the CFO based on the virtual subcarrier positions. When searching for wifi virtual subcarriers Google suggests: MIMO-OFDM for LTE, WiFi and WiMAX. Jun 14 '16 at 22:59