Update

[Asheesh Sharma]

Hello Sir,
Please find the updates below:

1) Script to trim web10glogs to sql-dbase. Status: completed Timeline: 10-01-2014

    Example database is attached.(web10gvars.db, web10gvars.html)

Pending:
1) Analysis and plotting scripts for web10g. Timeline: 10-01-2014 to 17-01-2014
2) Test to verify presence of local buffers in case of UDP packets 17-01-2014 to 25-01-2014

3)Lately, I found a test suite for android.
link: http://amorg.aut.bme.hu/projects/networkmonitor
I have requested for the source code, awaiting response.

Regards,
Asheesh Sharma

[Asheesh Sharma]

Sir,
Please find the updates below:

>>Analysis and plotting scripts for web10g. Deadline: 17-01-2014
I have created a plotting script to represent web10g variables (Vars Vs Timestamp). The script generates specified as well as all variable plots. Please suggest.
Attachments: plots.zip (Contents: 0_to_59.png, 59_to_119.png)

[Aaditeshwar Seth]

Couldn't make head of tail of these graphs. Arvind, are you supervising this?

Will be good to plot a variable such as congestion window or the roundtrip timer on the TCP sequence number graph, like what TCPtrace and other tools do. You should also attend my course on Internet measurements.

--
You received this message because you are subscribed to the Google Groups "ruralnet-act4d" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ruralnet-act4...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

-- 
Aaditeshwar Seth
Co-founder, Gram Vaani Community Media

http://gramvaani.org
http://www.linkedin.com/company/gram-vaani-community-media

[Arvind Mahla]

Ashish

As suggested you should plot for CurrCwnd (Current Congestion window), RTT, number of instances for slow start and Congestion Avoidance.

Arvind 

[Asheesh Sharma]

Sir,

Please find the updates below:

Plotting CurCwnd(congestion Window)-TimeSequence graph:
I have plotted congestion window w.r.t time-sequence graph. Please find the attachment. I have only plotted ACK events, others can be added.
Please suggest.

Regards,
Asheesh

[Aaditeshwar Seth]

Thanks. The next step is to plot these variables with the TCP sequence
numbers to understand long flow gaps. Additional data probably needs to
be collected from the rural sites, since the Delhi site was not
producing any flow gaps, and therefore web10g should be deployed on the
rural nodes and TCP data collected from them. Arvind: You need to
coordinate this

Rahul/Sameer: What's the update on the buffer behavior tests which had
to be redone with UDP?

[Arvind Mahla]

Sir

Web10g is already deployed at all rural nodes except one. The data (upto 22nd Dec) from all urban/rural nodes was checked for longflow gaps. Will again check for the gaps in the subsequent TCP data and report.

Arvind

On Sun, Jan 26, 2014 at 3:37 PM, Aaditeshwar Seth <as...@cse.iitd.ernet.in> wrote:

Thanks. The next step is to plot these variables with the TCP sequence numbers to understand long flow gaps. Additional data probably needs to be collected from the rural sites, since the Delhi site was not producing any flow gaps, and therefore web10g should be deployed on the rural nodes and TCP data collected from them. Arvind: You need to coordinate this

Rahul/Sameer: What's the update on the buffer behavior tests which had to be redone with UDP?

On 25-01-2014 00:07, Asheesh Sharma wrote:

Sir,

Please find the updates below:

Plotting CurCwnd(congestion Window)-TimeSequence graph:
I have plotted congestion window w.r.t time-sequence graph. Please find the attachment. I have only plotted ACK events, others can be added.
Please suggest.

Regards,
Asheesh
--
You received this message because you are subscribed to the Google Groups "ruralnet-act4d" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ruralnet-act4d+unsubscribe@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

--
Aaditeshwar Seth
Co-founder, Gram Vaani Community Media

http://gramvaani.org
http://www.linkedin.com/company/gram-vaani-community-media

--
You received this message because you are subscribed to the Google Groups "ruralnet-act4d" group.

To unsubscribe from this group and stop receiving emails from it, send an email to ruralnet-act4d+unsubscribe@googlegroups.com.

[Rahul Nishant]

Hi Sir,

Please find the updates below:

> Rahul/Sameer: What's the update on the buffer behavior tests which had
> to be redone with UDP?

- We have developed the UDP download test using iperf and we are testing
it. In the meantime we are analyzing the dumps collected from the
buffersize tests, from which we can get the one way latencies. The
problem with this latency is that, these one way latencies are when the
buffer is filled and we have packet losses. So, we are testing both the
Iperf and DITG(which we used for buffer test) to send packets at normal
rates without flooding the buffers.

- To evaluate the user experience in rural locations we have written a
script for extracting HAR files. We can simulate various browsers (
firefox, chrome, iOS etc) using Selenium Web drivers.

Thanks & regards,
Rahul & Sameer

[Aaditeshwar Seth]

I didn't understand the problem with capturing OTT. I think earlier you
guys did pilot tests and found NTP to be accurate to the order of a few
ms, so you should just have the measurement nodes and servers
synchronize with NTP.

Another solution, better solution actually, is to embed the sender
timestamp in the UDP message, and then at the receiver look at the
inter-dispatch (obtained from the timestamps in the UDP messages) and
inter-arrival times. You actually don't need absolute OTTs, relative
increase/decrease across consecutive packets will give you information
about the buffering that happens on the way.

[Asheesh Sharma]

Sir,
Please find attachment: Report.pdf

Regards,
Asheesh