Talk:Go-Back-N ARQ
From Wikipedia, the free encyclopedia
The article currently claims
- The sending window size must be no more than the number of sequence numbers to verify transmission in cases of any packet (any data or ACK packet) being dropped.
I think this is off-by-one -- it should say
- The sending window size must be smaller than the number of sequence numbers to verify transmission in cases of any packet (any data or ACK packet) being dropped.
Let me make up a very simple example:
- We use 4 sequence numbers (a 2-bit sequence number), and a transmitter window that can handle 4 packets.
- We transmit 1 character of payload per packet (unrealistic, but makes the example simpler)
- We are trying to transmit the message "Wikipedia".
The first 4 packets (0 1 2 3) get through fine. The receiver has sent back ACK0 ACK1 ACK2 and ACK3. The receiver has "Wiki" and is now expecting a packet with sequence number 0. Alas, ACK0 was destroyed in transit, so the transmitter re-sends packet 0 ("W"). If the ACK0 was *not* destroyed in transit, the transmitter would have sent a new packet 0 ("p").
The receiver now gets a W, and since its sequence number (0) is exactly what it was expecting for the 5th packet, it sends back ACK0 and adds it to its message so far: "WikiW".
If we change the system so that the transmitter window only holds 3 packets (or uses a sequence number that can count higher), then this problem doesn't occur.
--75.19.73.101 (talk) 19:05, 19 December 2007 (UTC)

