Transport Layer Overview & Protocols

The goal of the Transport Layer is to provide a reliable and efficient way for applications running on different hosts to communicate with each other.

Connectionless transport involves sending data segments without establishing a connection beforehand, while connection-oriented transport requires a connection setup process where both ends agree on communication parameters before data transfer.

<p>True (A)</p> Signup and view all the answers

<p>Source Port # = Identifies the specific application on the sending host Destination Port # = Identifies the specific application on the receiving host Sequence Number = Indicates the byte stream number of the first byte in the segment's data Acknowledgement Number = Indicates the byte stream number of the next byte expected from the other end Receive Window = Indicates the number of bytes the receiver is willing to accept Checksum = Used to check for errors in the segment's data Options Field = Provides additional information for connection management and congestion control</p> Signup and view all the answers

<p>TCP Slow Start gradually increases the sending rate of data in the beginning of a connection to avoid overwhelming networks and minimize the risk of congestion.</p> Signup and view all the answers

<p>True (A)</p> Signup and view all the answers

<p>TCP Reno uses a multiplicative decrease to half the congestion window when packet loss is detected by triple duplicate ACKs, while TCP Tahoe uses a multiplicative decrease to one maximum segment size (MSS) when loss is detected by a timeout.</p> Signup and view all the answers

<p>TCP CUBIC employs a faster initial ramp-up phase, increasing the congestion window based on the cube of the time difference to reach the target transmission rate, followed by a slower increase as it gets closer to the target.</p> Signup and view all the answers

<p>The bottleneck link is the most restrictive point in the network path, impacting the entire data flow. Understanding the bottleneck helps to identify the root cause of congestion and tailor congestion control mechanisms to optimize network performance.</p> Signup and view all the answers

<p>Minimizing the utilization of network resources (B)</p> Signup and view all the answers

<p>ECN is a network-assisted congestion control mechanism that enables routers to inform TCP senders about potential congestion in the network without relying on packet loss.</p> Signup and view all the answers

<p>The fairness goal in TCP congestion control aims to distribute network resources as much as possible to minimize bottlenecks and provide similar throughput to multiple active connections.</p> Signup and view all the answers

<p>Multimedia applications may choose UDP over TCP because it prioritizes minimizing latency and avoiding data loss, which are critical factors for real-time multimedia streaming. TCP's congestion control and reliability mechanisms can introduce delays that are undesirable for real-time applications.</p> Signup and view all the answers

<p>QUIC was developed to address performance limitations encountered with traditional TCP-based HTTP/2 communication, particularly in high-latency and packet-loss prone environments like wireless networks. QUIC aims to improve performance, reduce latency and offer a more efficient data transfer mechanism in these contexts.</p> Signup and view all the answers

<p>True (A)</p> Signup and view all the answers

The transport layer provides communication between application processes running on different hosts. It breaks application messages into segments, adds transport headers, and passes them to the network layer. On the receiver side, it reassembles segments into messages and delivers them to the application layer.

Multiplexing allows multiple application processes to share the same transport layer connection. This is achieved by adding transport headers to data, which contain information like source and destination port numbers, allowing the receiver to demultiplex the data to the correct application process.

Demultiplexing is the process of separating data received from a transport layer connection into the corresponding application processes based on the information in the transport header (e.g., port numbers).

UDP (User Datagram Protocol) is a connectionless, unreliable transport protocol. This means that data is sent as individual packets, with no connection setup, and there are no guarantees about delivery or order.

TCP (Transmission Control Protocol) is a connection-oriented, reliable transport protocol, providing in-order delivery and flow control. It uses a three-way handshake to establish a connection, and acknowledges received data, ensuring reliable data transfer.

A TCP segment carries information for data transport. This includes source and destination port numbers, sequence numbers for ordering data, acknowledgement numbers for reliable transfer, and other headers for flow and congestion control.

TCP sequence numbers are applied to each byte within the data stream. They provide a mechanism for the receiver to order data and detect packet loss.

TCP uses cumulative acknowledgements (ACKs). ACK(n) acknowledges all packets up to and including sequence number n.

TCP's reliable data transfer ensures packets arrive in order without loss. It utilizes sequence numbers, acknowledgements, timeouts, and retransmissions to achieve this reliability.

In TCP, the receiver window size specifies the amount of data the receiver is willing to accept. This helps to ensure that the receiver doesn't get overwhelmed with data and can process it efficiently.

TCP uses a three-way handshake to establish a connection. This process involves the sender and receiver exchanging synchronization (SYN) segments to ensure both sides are ready for data exchange.

TCP congestion control is a mechanism to prevent the network from being overloaded. It dynamically adjusts sending rates based on network conditions to avoid congestion and ensure efficient data transfer, preventing network crashes, dropped packets, and delays.

The Internet checksum is a simple method for error detection in packets. It involves treating packet contents as a sequence of 16-bit integers and adding them together, taking the one's complement sum to calculate the checksum. This can be used for detecting errors like bit flips during transmission.

rdt 2.0 uses acknowledgements (ACKs) and negative acknowledgements (NAKs) to handle bit errors. The receiver sends an ACK to acknowledge a correctly received packet or a NAK to signal an error, prompting the sender to retransmit the packet.

rdt 2.1 addresses a flaw in rdt 2.0 where ACK/NAK corruption could lead to unnecessary retransmissions. It adds sequence numbers to packets to distinguish duplicate transmissions and allows the receiver to discard duplicates.

rdt 2.2 is a NAK-free protocol that uses only ACKs for reliability. Instead of NAKs, the receiver sends an ACK for the last correctly received packet, including the sequence number of the acknowledged packet. This reduces the number of messages exchanged.

rdt3.0 introduces timeouts to handle packet loss. The sender waits for an ACK for a reasonable period of time and retransmits the packet if no ACK is received. The receiver must specify the sequence number of the acknowledged packet.

Sender utilization is a measure of the fraction of time the sender is actively sending data. In stop-and-wait protocols like rdt3.0, utilization is limited because the sender has to wait for an ACK before sending the next packet.

Pipelining allows the sender to send multiple packets before receiving acknowledgements. This can significantly improve sender utilization and network performance.

Go-Back-N is a reliable data transfer protocol that uses a window of up to N packets. The sender can transmit N consecutive packets before receiving an ACK. The receiver sends cumulative ACKs, acknowledging all packets up to a specific sequence number.

Selective repeat is a more efficient reliable data transfer protocol than Go-Back-N. It individually acknowledges each packet and allows out-of-order packets to be buffered. The sender maintains a timer for each unacknowledged packet and retransmits only lost packets.

TCP uses a sliding window to control data flow. The sender can only transmit data within the window size, which is adjusted based on network conditions like congestion and the receiver's ability to process data.

TCP utilizes a time-based timeout mechanism for detecting packet loss. The sender maintains a timer for each unacknowledged packet and retransmits the packet if no acknowledgement is received within a specific time interval.

TCP uses a 4-tuple for connection identification, which includes the source and destination IP addresses and port numbers. This ensures that data is delivered to the correct application process on the destination host.

EstimatedRTT (Estimated Round Trip Time) is an estimation of the round trip time for data transmission. It is calculated as an exponential weighted moving average of recent SampleRTTs (measured time from segment transmission to ACK receipt).

TCP's congestion control mechanisms are designed to prevent network overload. When congestion is detected, the sender decreases its transmission rate to reduce the load on the network. Conversely, it increases its transmission rate when congestion levels decrease.

The TCP slow start mechanism gradually increases the congestion window size during connection setup. This allows the sender to probe the network conditions and efficiently increase its transmission rate while avoiding congestion.

TCP implements a fast retransmit mechanism to quickly retransmit lost packets. When the receiver detects a packet loss, it sends duplicate acknowledgements (ACKs) to signal the loss. The sender responds by immediately retransmitting the lost packet, minimizing delay and improving performance.

TCP's fast recovery mechanism is triggered during congestion. When a packet loss is detected, the sender reduces the congestion window size, but it continues to transmit data at a reduced rate. This allows the sender to adjust to congestion without completely halting data transmission.