[HW] HW#04 2022

317 views
Skip to first unread message

夏秋如

unread,
Nov 7, 2022, 5:38:34 AM11/7/22
to NCCU Networking@MIS
Q1. TCP fast retransmission mechanism has no timer.
A1. False

Q2. In TCP, ACK number is the last byte received from the sender.
A2. False

Q3. In TCP, if an ack acknowledges an un-acked segment, the sender will always restart a timer.
A3. False

Q4. The sender’s view of ‘Selective Repeat’ is always not the same as the receiver’s view.
A4. False

Q5. In TCP, a client must terminate a connection before server.
A5. False

Q6. TCP offered load is always larger than sender’s transmission rate (including retransmission)
A6. False

Q7. For TCP congestion control, once a new ACK is received, it always goes to “congestion avoidance” state.
A7. False

Q8. It is possible that when timer timeouts, the GBN does not retransmit exactly N segments.
A8. True

Q9. Selective Repeat’s sender and receiver may have different ‘views’ of the pipeline; but the ‘view’ of GBN’s sender and receiver must be the same.
A9.False

夏秋如

unread,
Nov 7, 2022, 5:39:36 AM11/7/22
to NCCU Networking@MIS
同學如果對解答有疑問可以在這上面提問 謝謝!
夏秋如 在 2022年11月7日 星期一下午6:38:34 [UTC+8] 的信中寫道:

余則緯

unread,
Nov 7, 2022, 6:57:37 AM11/7/22
to NCCU Networking@MIS
助教好,想請教這兩題:
Screenshot 2022-11-07 195644.png
Screenshot 2022-11-07 195659.png
謝謝!

Prof. Hsiao

unread,
Nov 7, 2022, 7:25:39 AM11/7/22
to NCCU Networking@MIS
Q3. In TCP, if an ack acknowledges an un-acked segment, the sender will always restart a timer.
A3. False

投影片 3-65
收到 ACK 不一定要開 timer。
FSM 中左下角的事件有一個判斷式,
若 window 內仍有 in-flight 封包才需要開 timer。


Q9. Selective Repeat’s sender and receiver may have different ‘views’ of the pipeline; but the ‘view’ of GBN’s sender and receiver must be the same.
A9.False

不管是任何一種 pipeline 機制,
只要有 in-flight 的封包,
sender 和 receiver 的 window "view" 應該會長得不一樣。
都有可能發生一方已經移動 window,但是另外一方還未移動 window 的狀況。

Fang Yu Hsing

unread,
Nov 7, 2022, 7:42:35 AM11/7/22
to NCCU Networking@MIS
老師您好,
我想要請問這一題~
印象中gbn在其中一個timer timeout時會重新傳全部n個segment
好像跟題目敘述不符。
謝謝!
Notes_221107_203955_d20.jpg
chiu3...@gmail.com 在 2022年11月7日 星期一下午6:38:34 [UTC+8] 的信中寫道:

余則緯

unread,
Nov 7, 2022, 7:52:27 AM11/7/22
to NCCU Networking@MIS
真的很感謝老師即時的回覆,非常有幫助!

余則緯

unread,
Nov 7, 2022, 7:53:31 AM11/7/22
to NCCU Networking@MIS
這我我是想有可能是in-flight的封包數小於N,那go back的時候就不會到N那麼多,不知道是不是這樣

Prof. Hsiao

unread,
Nov 7, 2022, 8:10:04 AM11/7/22
to NCCU Networking@MIS
Q8. It is possible that when timer timeouts, the GBN does not retransmit exactly N segments.
A8. True

Window 裡面也有機會不是全部都黃色 (in-flight) 的封包,
也有可能有一些藍色未送出的封包。

投影片 3-107
注意重傳的時候是 base 到 nextseqnum -1,
這有可能有藍色的未送出封包。
我們只是說有很大的機會 window 若是全部黃色的,
會是 N 個,因此命名為 GBN。
但不總是是 N 個。


Fang Yu Hsing

unread,
Nov 7, 2022, 8:18:00 AM11/7/22
to NCCU Networking@MIS
了解,謝謝老師~

Prof. Hsiao 在 2022年11月7日 星期一晚上9:10:04 [UTC+8] 的信中寫道:

Bryant Tian

unread,
Nov 7, 2022, 9:50:17 PM11/7/22
to NCCU Networking@MIS
不好意思 想請問第一題 server在收到三個重複的ACK不是就會retransmit了嗎?這樣應該就沒用到timer了吧?

chiu3...@gmail.com 在 2022年11月7日 星期一下午6:38:34 [UTC+8] 的信中寫道:
Q1. TCP fast retransmission mechanism has no timer.

Prof. Hsiao

unread,
Nov 7, 2022, 10:20:48 PM11/7/22
to NCCU Networking@MIS
Fast retransmission 這個機制是因為可以比較快速的根據 3 duplicate ACK 而提早重送,
但這並不表示 timer 的機制就完全消失了。
你可以假想萬一一直只有 2 duplicate ACK,
而第三個 duplicate ACK 一直沒收到。
那麼在 sender 端還是最終需要 timer 的機制來重送。



wu06...@gmail.com 在 2022年11月8日 星期二上午10:50:17 [UTC+8] 的信中寫道:
不好意思 想請問第一題 server在收到三個重複的ACK不是就會retransmit了嗎?這樣應該就沒用到timer了吧?

Prof. Hsiao

unread,
Nov 7, 2022, 10:21:57 PM11/7/22
to NCCU Networking@MIS
在投影片 3-68 裡面仍舊是有 timer 在倒數的。

Prof. Hsiao 在 2022年11月8日 星期二上午11:20:48 [UTC+8] 的信中寫道:
Reply all
Reply to author
Forward
0 new messages