1. The following two questions are related to congestion control in wired networks. Let us assume
that the size of buffer is infinite.
a. Suppose "token bucket" mechanism for congestion control. Explain how to compute a queuing delay (i.e., from when a packet comes in the bucket to when it goes out). Note
that, you do not need to use mathematical equations.
b. Suppose "dynamic window management" for congestion control. Explain what problems occur under the large window size on connection created.
In IEEE 802.11, a node retransmits a packet if the previous transmission of the packet is unsuccessful. If TCP over 802.11 is used, both protocols may generate a large number of retransmissions of a given packet to ensure its reliable transmission. Explain why the
retransmissions are need for both TCP and 802.11 even though these have the same purpose.
2. The EDCF (Enhanced DCF) in IEEE 802.11e maintains 4 different queues to support access categories (AC) (See the sides). IEEE 802.11e describes only MAC and PHY layers. Even though the upper layer (e.g., IP layer) can categorize the characteristics of 4 different packets
(Voice, Video, Best Effort, and Background), the MAC layer does not check the category due to encapsulation. Explain how the EDCF identifies 4 different AC of packets from the upper layer without the reference of IP or TCP headers.
3. The following question is related to the SPEED routing protocol:
Multiple speeds are computed in the MAC layer. A forwarder with the highest speed is selected on MAC layer feedback. This selection may not produce a shortest path. Explain why SPEED uses this approach instead of other shortest or reliable path selections and the benefit of this approach against the shortest and reliable paths.