WARP Project Forums - Wireless Open-Access Research Platform

You are not logged in.

#1 2017-Jun-23 11:38:13

ShuaiWWW
Member
Registered: 2017-Jun-23
Posts: 1

Why does packet detection choose 16 as length of correlation window?

As we know, Schmidl&Cox is a popular OFDM preamble detection approach. Half length of STS is the default length of auto-correlation window in Schmidl&Cox. However, 802.11 reference design choose 16 instead of 80 (160/2) as the windows length. I am wondering what are the benefits of this?

Offline

 

#2 2017-Jun-23 14:20:01

murphpo
Administrator
From: Mango Communications
Registered: 2006-Jul-03
Posts: 5159

Re: Why does packet detection choose 16 as length of correlation window?

The core idea of the auto-correlation packet detection block is to search for the large auto-correlation in the periodic STF. The STF is composed of 10 copies of a 16-sample sequence (10 STS). This periodicity defines the autocorrelation length. The Schmidl-Cox technique normalizes the raw auto-correlation value with the magnitude of the same samples, in effect measuring how "correlate-y" the samples are independent of the Rx power. This comparison is implemented as the correlation threshold in our designs. Our logic also includes a minimum magnitude threshold to reject the (many) false positives that can occur with very small signals in the time between actual transmissions. You could use a longer correlation window, but this would delay the detection decision which reduces the time AGC has to settle. We aim for a quick packet detection, then refine the timing using the LTF correlator (or reject the detection event if there is no subsequent LTF correlation event).

Offline

 

Board footer