The packet detection algorithm is implemented as a double sliding
window as described in OFDM Wireless LANs [1], Chapter 2. The autocorrelation of L-STF short training symbols is used to
return an estimated packet-start offset. In a robust system, the next
stage will refine this estimate with symbol timing detection using
the L-LTF.

As shown in the figure, the received signal, *r*_{n},
is delayed then correlated in two sliding windows independently. The
packet detection processing output provides decision statistics (*m*_{n})
of the received waveform.

Window *C* autocorrelates between
the received signal and the delayed version, *c*_{n}.

Window *P* calculates the energy
received in the autocorrelation window, *p*_{n}.

The decision statistics, *m*_{n},
normalize the autocorrelation by *p*_{n} so
that the decision statistic is not dependent on the absolute received
power level.

The decision statistics
provide visual information resulting from the autocorrelation process
that is useful when selecting the appropriate threshold value for
the input waveform. The recommended default value of 0.5 for `threshold`

favors
false detections over missed detections considering a range of SNRs
and various antenna configurations.

In the sliding window calculations, *D* is
the period of the L-STF short training
symbols and *N*_{R} is the number
of receive antennas.

Packet detection processing follows this flow chart:

*L*_{STF_SYMBOL} is the
length of an L-STF symbol.

### Note

This function supports packet detection of OFDM modulated signals
only.