Message-ID: <525CEE98.3010300@intec.ugent.be> (sfid-20131015_093651_993761_CABDE8DE)
Date: Tue, 15 Oct 2013 09:28:24 +0200
From: Floris Van den Abeele <floris.vandenabeele@intec.ugent.be>
MIME-Version: 1.0
To: linux-wireless@vger.kernel.org
Subject: Lower throughput in Ad-hoc than in Infrastructure mode?
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Sender: linux-wireless-owner@vger.kernel.org

While running throughput experiments using iperf I noticed that UDP
throughput in ad-hoc mode was considerably lower than in Infrastructure
mode. The setups I'm comparing is a 802.11a ad-hoc network consisting of
two clients vs a 802.11a infrastructure network consisting of an access
point and a client (11a because the tests were run in the 5GHz band).

For ad-hoc mode I'm consistenly seeing a lower throughput than for
infrastructure mode: i.e. 22.4 Mbits/sec vs 28.2 Mbits/sec. Note that
while sniffing the medium radiotap headers report a PHY data rate of
54Mbit/s in both cases. There is no other traffic on the Wi-Fi channel
in question (44). Considering the medium access for both setups is
similair I would expect to find similair throughputs. Is this a known
issue with the linux wireless/mac80211/ath9k driver stack or is this
behaviour conform the 802.11 standard (based on my knowledge I would say
no to the latter)? My test specs are below.

For testing I'm using Sparklan WPEA-111N NIC's (AR9280). The nodes are
running OpenWrt attitude adjusment on r32482 (from 2012-06-21) with the
REGD patch enabled. I believe that this build of OpenWRT uses the
compat-wireless-2012-06-14.tar.bz2 drivers.

Kind regards,
Floris

-- 
Floris Van den Abeele
Department of Information Technology
Internet Based Communication Networks and Services (IBCN)
Ghent University - iMinds
Gaston Crommenlaan 8 (Bus 201), B-9050 Gent, Belgium
M: +32 (0)494 24 92 76
E: floris.vandenabeele@intec.UGent.be
W : www.ibcn.intec.UGent.be