On Tue, Apr 19, 2011 at 4:03 AM, Yeoh Chun Yeow <yeohchuny...@gmail.com> wrote: > I have setup a similar network topology using dual radio. > Mesh A1 < ----- Channel 44 -------> Mesh A1 bridge Mesh B1 <----- Channel 40 > ------> Mesh B2 > Throughput from Mesh A1 to Mesh B2 is 10.5Mbps. > Throughput from Mesh A1 to Mesh A1 is 17.6Mbps. > I think that throughput is degraded due to the bridging between Mesh A1 to > Mesh B1. But it is not cut by half. Just happen to be your setup.
A couple of comments on this: - If you are planning on deploying a multi channel mesh, it will pay off very quickly to extend open80211s to support it and not do bridging. Bridging is intended for heterogeneous intefaces (e.g. mesh and AP, or mesh and ethernet). But it adds additional overhead (an additional address lookup per frame, larger mesh headers, etc.). - To further investigate the lower throughput you observe, you should look at sniffer captures. Are mesh frames transmitted at the expected data rate? How busy is the channel? Are there many retries/losses? Cheers and thanks for sharing your results with the list. Javier > Mesh A1 <------ Channel 44 ---------> Mesh A2 <---------- Channel 44 > --------> Mesh A3 > Throughput from Mesh A1 to Mesh A2 is 17.8Mbps. > Throughput from Mesh A1 to Mesh A3 (next hop is Mesh A2) is 7.68Mbps. > Sharing the same radio, then it is cut by half. > On Tue, Apr 19, 2011 at 11:58 AM, Yeoh Chun Yeow <yeohchuny...@gmail.com> > wrote: >> >> Try use channel 6 instead of channel 1. >> >> On Tue, Apr 19, 2011 at 11:25 AM, Zaki <zaki...@gmail.com> wrote: >>> >>> Hi Xianghua, >>> >>> Yes, it is cut to half. Tested using wired between the WLAN cards, it >>> improved up no loss in bitrate, provided the channel are far between each >>> other (Channel 1 with Channel 11), otherwise the degradation is observed >>> too. In wireless, we did try Channel 1 with Channel 36, it was good, no >>> bitrate loss observed but it is inconsistent, maybe by luck. So, there is >>> no conclusion on this yet at the moment. Study is on-going. Could it be >>> due to the bridge used?. >>> >>> Have you encountered this problem and it is solved by batman-adv >>> before?. Appreciate your inputs on this. >>> >>> Cheers, >>> Zaki. >>> >>> >>> >>> On Sat, Apr 16, 2011 at 12:38 PM, Xianghua Xiao <xiaoxiang...@gmail.com> >>> wrote: >>>> >>>> it puzzles me that with dual radio on your PC2 still the throughput is >>>> cut in half? >>>> maybe you can try batman-adv on PC2 >>>> >>>> On Fri, Apr 15, 2011 at 11:56 AM, Javier Cardona <jav...@cozybit.com> >>>> wrote: >>>> > Zaki, >>>> > >>>> > On Fri, Apr 15, 2011 at 1:46 AM, Zaki <zaki...@gmail.com> wrote: >>>> >> As far as the kernel, configurations and environment, we conducted >>>> >> the tests >>>> >> with the same setup. We follow exactly like what is mentioned in the >>>> >> HOWTO. Eventhough the environment is not 'quiet', we did our test by >>>> >> using >>>> >> conducted testing. >>>> >> >>>> >> Open MESH results are okay (the Rx Signal Strength is about -10 dBm >>>> >> consistently). >>>> >> >>>> >> For SAE, we noticed that the Rx signal strength fluctuated, from 2dBm >>>> >> until >>>> >> -70dBm. It's a huge fluctuation there. It was observed on all WLAN >>>> >> cards >>>> >> that we used (all based on Atheros chipset). That resulted into the >>>> >> bitrate >>>> >> seen fluctuated as well, at times didn't make it to the endpoint. >>>> >> >>>> >> So, we are wondering what would make the fluctuation to happen. Is >>>> >> it >>>> >> something to do with the wpa_supplicant <-> driver relationship or >>>> >> anything >>>> >> else that could be the real disturbance?. Appreciate your input. >>>> >> Thanks. >>>> > >>>> > Ah, good point. Can you post your wpa_supplicant.conf (without the >>>> > commented lines) here? >>>> > wpa_supplicant's default configuration scans periodically for new >>>> > networks. If you are only using it for mesh, this is not needed and >>>> > can be disabled with ap_scan=0 >>>> > >>>> > Alternatively, to rule out wpa_supplicant issues, you can try the >>>> > simpler authsae daemon: https://github.com/cozybit/authsae >>>> > >>>> > Cheers, >>>> > >>>> > Javier >>>> > >>>> >> On Thu, Apr 14, 2011 at 12:06 PM, Javier Cardona <jav...@cozybit.com> >>>> >> wrote: >>>> >>> >>>> >>> Hi Zaki, >>>> >>> >>>> >>> On Tue, Apr 12, 2011 at 11:29 PM, Zaki <zaki...@gmail.com> wrote: >>>> >>> > We are actively doing testings with Open and SAE Mesh. While Open >>>> >>> > Mesh's >>>> >>> > throughput looks fine, SAE's throughput looks a bit strange. >>>> >>> > Please >>>> >>> > refer >>>> >>> > to here http://www.ahmadzaki.com/?p=16 to see the block diagram of >>>> >>> > the >>>> >>> > setup >>>> >>> > we are using and the detail results out of them. Anybody know if >>>> >>> > the >>>> >>> > results are normal?. >>>> >>> >>>> >>> SAE is only used for authentication. Once a station is >>>> >>> authenticated, >>>> >>> everything should behave exactly the same as an open mesh, so there >>>> >>> should be no effect on throughput (this may change once AMPE and >>>> >>> encryption are released). I would suspect on other factors, for >>>> >>> instance configuration, kernel or changes to your wireless >>>> >>> environment >>>> >>> (are you running your tests in a controlled environment?). To be >>>> >>> absolutely sure, you could repeat your SAE throughput measurements >>>> >>> with the same kernel and configuration but over unauthenticated peer >>>> >>> links. The HOWTO explains how to create unauthenticated peer links >>>> >>> (http://o11s.org/trac/wiki/HOWTO#openmesh). Throughput numbers >>>> >>> should >>>> >>> be identical. >>>> >>> >>>> >>> Cheers, >>>> >>> >>>> >>> Javier >>>> >>> >>>> >>> -- >>>> >>> Javier Cardona >>>> >>> cozybit Inc. >>>> >>> http://www.cozybit.com >>>> >> >>>> >> >>>> > >>>> > >>>> > >>>> > -- >>>> > Javier Cardona >>>> > cozybit Inc. >>>> > http://www.cozybit.com >>>> > _______________________________________________ >>>> > Devel mailing list >>>> > Devel@lists.open80211s.org >>>> > http://open80211s.com/mailman/listinfo/devel >>>> > >>>> _______________________________________________ >>>> Devel mailing list >>>> Devel@lists.open80211s.org >>>> http://open80211s.com/mailman/listinfo/devel >>> >>> >>> _______________________________________________ >>> Devel mailing list >>> Devel@lists.open80211s.org >>> http://open80211s.com/mailman/listinfo/devel >>> >> > > -- Javier Cardona cozybit Inc. http://www.cozybit.com _______________________________________________ Devel mailing list Devel@lists.open80211s.org http://open80211s.com/mailman/listinfo/devel
