video was Re: ideas was Re: [CUWiN] Activity?

[Tom Sparks]

the codec I am looking at using are:
Theora/Vorbis for the videos streams:
frame size 352x288 25fps 150 KB / audio 44100 Hz
Stereo 128KB
frame size 640x480 25fps 512 KB / audio 44100 Hz
Stereo 128KB
(may build custom video cameras)

Vorbis for the audio stream:
audio 44100 Hz Stereo 128KB

Speex for VoIP
I have not done more as this is not high on the list
but it well be used everywhere were textural based
message can be
entered eg: forums

there is a pdf called "VIDEO STREAMING OVER WIRELESS
NETWORKS" by
Xiaoqing Zhu and Bernd Girod that may be worth a read

On Thu, 2008-08-07 at 13:05 -0400, Bob Keyes wrote:
> Hello Toms and CuWIN,
> Right now I have a paid job making a mesh network
which can handle video. 
> I've signed an NDA and there will be certain parts
of this system which 
> they will want to keep proprietary. But there are
some other parts of the 
> project in which we can work together.
> 
> Performance testing is one of these areas. Video and
VoIP are tested now 
> with 'it seems okay', end-user experience. But this
is hardly good 
> science. I am trying to come up with measurements
which we can compare to 
> the competition and our own product revisions. iperf
just won't cut it. 
> We're looking into ixchariot but I doubt this will
be able to provide a 
> definitive measurement either, though it's probably
better than iperf 
> because it is so flexible.
> 
> The other things I can offer to help you is a little
bit of knowledge I 
> had before I came to work on this project: QoS and
so forth is important, 
> but there's much more than what's going on in the IP
layer than can effect 
> the quality of video on the mesh network. Like VoIP,
latency is important, 
> but some packets can be lost without much of a
degradation of quality. 
> It's important to learn the limitations of delay and
loss in the codec you 
> are concerned with, and to design the network around
them. Designing a 
> network for such applications requires more than
just installing some 
> generic mesh nodes on some convenient buildings and
hoping for the best. 
> If you are lucky enough that the nodes are close
together, and that 
> there's not much other 2.4 ghz activity, it can work
- but what about when 
> it stops working or is degraded? Do you have the
knowledge, skills, and 
> tools to figure out what went wrong? If you're going
to be selling nodes 
> for little money as a consumer devices, and telling
people they can handle 
> video, don't over-promise on performance or
reliability.
> 
> Regards,
> Bob Keyes
> 



      Win a MacBook Air or iPod touch with Yahoo!7. http://au.docs.yahoo.com/homepageset