ORBIT-USER: Analyzing Result

[Zhibin Wu]

I think the problems is:
the user tries to do two different things with one TIME:SUM filter:
calculating throughput and determining packet loss ratio per second,
which is impossible.
Think this way, given the SQL table of receiver_port, can we use a
single SQL query to get the above two measurements. I do not think so.
I think either design a new filter to hint some packet loss (e.g  a
filter to report the occurrence of "NONCONSECUTIVE metric"), or the
user shall still use SAMPLE based filter and compare two tables
themselves with some post-processing script.
And I still prefer the latter, because it is the right way to measure
accurate packet loss ratio.

Zhibin

On 1/29/07, Pandurang Kamat <pkamat at winlab.rutgers.edu> wrote:
> This is probably because the receiver itself and/or the measurement
> thread could have started a trifle later than the sender threads. If you
> notice that the rest of the rows have pretty consistent number of
> packets (36 or 37) in every second in both the tables (I checked for a
> few rows).
> Again, it may be that in one sender row it shows sum of 37 pkts but on
> the rcv side you see sum of 36 pkts. This is because of the granularity
> of timers and is to be expected; the missing packet shows up in the next
> row. You should be looking at average data in this expt and not
> necessarily individual values.
>
> -Pandurang
>
> Arinto Murdopo wrote:
> > Dear All
> >
> > I have question regarding analyzing result after conducting expriment in
> > ORBIT testbed, based in this article
> > (http://www.orbit-lab.org/wiki/Documentation/OTG/UserManual
> > <http://www.orbit-lab.org/wiki/Documentation/OTG/UserManual>) , we can
> > measure the packet loss by comparing *pkt_seqno* columns between two
> > resulting tables, the receiverport table and senderport table.
> >
> > Suppose at time t =1, the pkt_seqno for those two tables is equals to 1
> > and at time t = 2 (1 second later) the pkt_seqno  for receiverport table
> > is 25 and for senderport table is 37 (as stated in the file that I
> > attach). How to interpret this results? Does it mean that we have 12
> > lost packets during t = 1 and t = 2?
> >
> > Here I also attach the result in excel format and also the script that I
> > use in this experiment. Actually what I did at that time is change the
> > RSSI value and also the percentage of packet received in run time using
> > iwpriv utility.
> >
> > Thank you
> >
> > Regards,
> >
> > Arinto
> >
> >
>


-- 
Zhibin Wu
Graduate Research Assistant
WINLAB, Rutgers University
Technology Centre of New Jersey
671 Route 1 South
North Brunswick, N.J. 08902-3390
Phone: 732-932-6857 x653
Fax:   732-932-6882