Chapter 5 2. Consider a simple UDP-based protocol for requesting files (based somewhat loosely on the Trivial File Transport Protocol, or TFTP). The client sends an initial file request, and the server answers (if the file can be sent) with the first data packet. Client and server then continue with a stop-and-wait transmission mechanism. (a) Describe a scenario by which a client might request one file but get another; you may allow the client application to exit abruptly and be restarted with the same port. (b) Propose a change in the protocol that will make this situation much less likely. 44. Consider a client and server using an RPC mechanism that includes a channel abstraction and boot IDs. (a) Give a scenario involving server reboot in which an RPC request is sent twice by the client and is executed twice by the server, with only a single ACK. (b) How might the client become aware this had happened? Would the client be sure it had happened?

Chapter 6 2. TCP uses a host-centric, feedback-based, window-based resource allocation model. How might TCP have been designed to use instead the following models? (a) Host-centric, feedback-based, and rate-based. (b) Router-centric and feedback-based. 9. Give an example of how non-preemption in the implementation of fair queuing leads to a different packet transmission order from bit-by-bit round-robin service. 30. Suppose two TCP connections share a path through a router R. The router's queue size is six segments; each connection has a stable congestion window of three segments. No congestion control is used by these connections. A third TCP connection now is attempted, also through R. The third connection does not use congestion control either. Describe a scenario in which, for at least a while, the third connection gets none of the available bandwidth, and the first two connections proceed with 50% each. Does it matter if the third connection uses slow start? How does full congestion avoidance on the part of the first two connections help solve this? 44. Many real-time video applications run over UDP rather than TCP because they cannot tolerate

retransmission delays. However, this means video applications are not constrained by TCP's congestion- control algorithm. What impact does this have on TCP traffic? Be specific about the consequences.

Fortunately, these video applications often use RTP, which results in RTCP receiver reports being sent from the sink back to the source. These reports are sent periodically (e.g., once a second) and include the percentage of packets successfully received in the last reporting period. Describe how the source might use this information to adjust its rate in a TCP-compatible way.

Chapter 8 4. A good cryptographic hashing algorithm should produce random outputs; that is, the probability of any given hash value should be approximately the same as any other for randomly chosen input data. What would be the consequence of using a hash algorithm whose outputs were not random? Consider, for example, the case where some hash values are twice as likely to occur as others. 18. Suppose you want your firewall to block all incoming Telnet connections but to allow outbound Telnet connections. One approach would be to block all inbound packets to the designated Telnet port (23). (a) We might want to block inbound packets to other ports as well, but what inbound TCP connections must be permitted in order not to interfere with outbound Telnet? (b) Now suppose your firewall is allowed to use the TCP header Flags bits in addition to the port numbers. Explain how you can achieve the desired Telnet effect here while at the same time allowing no inbound TCP connections. 21. Why might an Internet Service Provider want to block certain outbound traffic?