מספר שאלות

1 view
Skip to first unread message

OK

unread,
Apr 12, 2010, 8:11:08 AM4/12/10
to 89350
שלום רב,

מספר שאלות לא ברורות -
1) בשאלות החזרה - שאלה 2, סעיף 2, משתמשים בגירסת HTTP 1.0 ומחשבים כמה
זמן זה ייקח. מה שלא הבנתי זה למה לא מכפילים את ה-RTT בלוג 12.5?
2) במועד א/ של שנת 2008 שאלה 2 -
כיצד מחשבים כמה slots ידרשו בממוצע להשלמת שידור של הודעה, שאורכה 2
מסגרות? 6 מסגרות?
3) במועד א/ השנה שאלה 3 סעיף ג -
למה מכפילים ב-10 את הביטוי (לא הבנתי למה לא מכפילים בזמן שליחת כל
החבילות).

בתודה מראש,

אור

yaron winter

unread,
Apr 12, 2010, 11:36:00 AM4/12/10
to 89...@googlegroups.com
Hi OK,

(1) It is a typo mistake - the propagation delay should had been multiplied by log(12.5) as well.

(2) Two nodes: x & y.

- the probability that x will successfully transmit in a given slot = p*(1-p)
- the probability that either x or y will successfully transmit = 2*p*(1-p), which is also the network utilization percentage.

So for maximizing utilization we derive the function f(p) = 2p(1-p) and get: optimal p = 0.5.

Now, suppose x wants to transmits message of length 2 slots.
Let's denote by A the probability that x successfully transmits in slot.
Let's denote by P(X=k) the probability that it takes k(>=1) slots until x transmits successfully a single packet (i.e slot).
Thus, the average number of slots requires for x to transmit the first packet is:

E(X) = 1/A

And the average number of slots for transmitting N packets is simply N*E(X)

since A=0.25 (see above), it takes 8 slots to transmit 2 packets and 24 slots for transmitting 6 packets.

(3) Here we compute the average number of packets loss.

Let's denote by T the accumulated transmission time of a packet (namely: A's + B's + C's transmission time) and by TO the time-out.
Let's denote by X the number of packets loss.

So the average time requires for delivering the first packet (1400B) is:

T + (T+TO)*E(X)

This is why 10 (the time out) is multiplied there (actually they forgot to add the transmission time, but any lost packet must be retransmitted).

I hope it helped you.
Yaron


2010/4/12 OK <orka...@gmail.com>

--
You received this message because you are subscribed to the Google Groups "89350" group.
To post to this group, send email to 89...@googlegroups.com.
To unsubscribe from this group, send email to 89350+un...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/89350?hl=en.


OK

unread,
Apr 12, 2010, 1:30:38 PM4/12/10
to 89350
יש קצת בעיה ב-2 כי אתה רוצה שהסלוטים יהיו רצופים ולכן צריך לא להכפיל ב-
N אלא להעלות בחזקה. יש עוד אופציות?

> 2010/4/12 OK <orkam...@gmail.com>


>
>
>
> > שלום רב,
>
> > מספר שאלות לא ברורות -
> > 1) בשאלות החזרה - שאלה 2, סעיף 2, משתמשים בגירסת HTTP 1.0 ומחשבים כמה
> > זמן זה ייקח. מה שלא הבנתי זה למה לא מכפילים את ה-RTT בלוג 12.5?
> > 2) במועד א/ של שנת 2008 שאלה 2 -
> > כיצד מחשבים כמה slots ידרשו בממוצע להשלמת שידור של הודעה, שאורכה 2
> > מסגרות? 6 מסגרות?
> > 3) במועד א/ השנה שאלה 3 סעיף ג -
> > למה מכפילים ב-10 את הביטוי (לא הבנתי למה לא מכפילים בזמן שליחת כל
> > החבילות).
>
> > בתודה מראש,
>
> > אור
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "89350" group.
> > To post to this group, send email to 89...@googlegroups.com.
> > To unsubscribe from this group, send email to

> > 89350+un...@googlegroups.com <89350%2Bunsu...@googlegroups.com>.

yaron winter

unread,
Apr 12, 2010, 3:32:01 PM4/12/10
to 89...@googlegroups.com
Are you sure that there must be two consecutive slots? Can't a frame of length n slots be transmitted packet by packet?

Anyway, if there must be enough consecutive slots it changes things only slightly:

A=p*(1-p)^n is the probability that x will get n consecutive slots

Thus:

P(X=k+1) = A*(1-A)^k is the probability that x will get its n consecutive  slots in the k+1 attempt (slot).
And again: E(X) = 1/A

A(2) = 0.125 ==> 8+1 slots for message of length 2
A(6) = 1/128 ==> 128+5 slots for message of length 6

Yaron


To unsubscribe from this group, send email to 89350+un...@googlegroups.com.

yaron winter

unread,
Apr 13, 2010, 10:32:06 AM4/13/10
to 89...@googlegroups.com
Hi OK,

Take a look at the attached file, which contains the solutions of Moed Alef 2008 as published by the stuff.
According to this solution a frame of length n packets can indeed be transmitted packet by packet.

Yaron
2007-8_A_Answ.doc
Reply all
Reply to author
Forward
0 new messages