Chapter 2 Point-to-Point Data Transfer

Question 1

In the data link layer what is a “frame”

Answer 1

A unit of data exchanged between nodes

Question 2

Two main ways of handling corrupted frames

Answer 2

Forward Error Correction: Receiver corrects errors based on added redundancy
Error Dectection: Receiver identifies erros and dsicards corrupted frames

Question 3

Stop-and-wait Protocol

Answer 3

Send one frame, wait for an ACK before sending next, requires sequence numbers for frame recognition and timeout mechanisms to avoid deadlock.

Question 4

Why can stop-and-wait protocol be inefficient?

Answer 4

Long propagation delays relative to transmission times.

Question 5

What are two basic pipelining strategies in reliable data transfer

Answer 5

Go-Back-N sliding window
Selective Repeat sliding window

Question 6

Go-Back-N sliding window

Answer 6

The sender can transmit multiple frames, the receiver does NOT buffer out-of-oder frames upon detecting a lost frame. Sender retransmits that frame plus all subsequent frames.

Question 7

What are cumulative Acks? Where is it used?

Answer 7

These are acks that are all the frames transmitted before frame i-1. Used in Go-Back-N

Question 8

What is a sending window?

Answer 8

Consists of a sequence numbers of those frames that have been trasnsmitted but not yet acknowledged. As well as new frames the sender is permitted to send. The window slides through this sequence numbers.

Question 9

Why must Go-Back-n limit the sending window size when using small sequence numbers?

Answer 9

To avoid ambiguity between new and retransmitted frames due to sequence number reuse

Question 10

Maximum sending window size with k-bit sequencenumber in Go-Back-N?

Answer 10

2^k - 1

Question 11

Why can’t Go-Back-n use a window size equal to the number of sequence numbers?

Answer 11

Because if all ACK’s are lost, retransmissions may appear to the receiver as new frames due to sequence number overlap

Question 12

What is the bandwidth-delay product?

Answer 12

Bandwidth * Round-trip Time; represents amount of data that can be in the minimum window size needed for full channel utilization.

Question 13

Selective Repeat Sliding Window

Answer 13

Unlike go-back-n protocol, the receiver can now buffer out-of-order frames until the missing frames have been recieved. The sender only retransmits frames thought to have NOT be recevied correctly.
- uses a cumulative ack
- reciever maintains a window

Question 14

How does the sender in Selective Repeat decide when to retransmit?

Answer 14

It sets a timeout and only retransmits the oldest unACK’d frame when the timeout expires.

Question 15

Why add redundancy in data transmission

Answer 15

To enable error detection, by adding extra bits to the original data.

Question 16

Block Code?

Answer 16

A code that maps m message bits to an n = m + r bit codeword only using the current m bits.

Question 17

Systematic Code?

Answer 17

Input data is embedded on the output

Question 18

How is code rate defined

Answer 18

Code Rate = m/n

Question 19

Hamming Distance between two codewords?

Answer 19

Number of bit postions in which the two codewords differe

Question 20

Hamming Distance of a code?

Answer 20

THe minimum hamming distance among all valid codeword pairs.

Question 21

To DETECT up to “d” errors what hamming distance is needed?

Answer 21

Atleast d+1

Question 22

to CORRECT up to d errors, what hamming distance is needed?

Answer 22

At least 2d + 1

Question 23

How does the number of redundancy bits “r” scale with message length “m” and errors “d”?

Answer 23

r scales roughly as d log m for reliable correction

Question 24

Cyclic Redundancy Check

Answer 24

Error-detching code widely used at the link layer. It generates a checksum

Question 25

How is CRC Codeword formed

Answer 25

Append r redundancy bits to m messages so that n = m + r bits that are divisble under XOR

Question 26

How does a receiver use CRC to detect errors

Answer 26

It checks if the received codeword is divisble by G using XOR divison, if not than an error is detected.

Question 27

Convolutional Codes

Answer 27

Codes that output encoded bits based on current and previous message bits.

Question 28

Code rate of convolutional code

Answer 28

The ratio of message bits to codeword bits

Question 29

Difference between soft-decision and hard-decision decoding

Answer 29

Hard decision: Each bit is either 0 or 1 Soft-decision: uses physical layer info baout the signal condience levels to improve accuracy.

Question 30

4 methods for Frame Delimiting

Answer 30

-Flag bytes with byte stuffing -bit stuffing -coding violations -preamble + legnth field

Question 31

Framing in the link layer

Answer 31

the process of encapsulating data received from the network layer into smaller, manageable units called frames before transmission over the physical medium

Question 32

What fields helps the receiver identify how to process the frame payload?

Answer 32

A type field indicating the payloads protocol (IP)

Question 33

What happens when the sender transmits faster than the receiver can process?

Answer 33

Reciever buffer overflow or data loss.

Question 34

How do we deal with potential scenario of sender transmitting data faster than the receiver can deal with it?

Answer 34

Simply discard traffic Use a flow control mechanism.

Question 35

What is Converged Enchanced Ethernet?

Answer 35

an enchanced version of ethernet used in data centers to support advanced control mechanisms

Question 36

Priority flow control

Answer 36

A machanism to send pause/resume frames for individual traffic

Question 37

The problem with flow control at the link layer?

Answer 37

It can have a cascading pause effect, a paused node may need to pause upstream nodes casuing congestion

Question 38

What is QCN? Quantized Congestion Notification

Answer 38

A congestion management technique in CEEE that helps control the cascading pause problem for long lived flows. Used in high performance data centers

Question 39

End-to-end argument say about protocol design

Answer 39

That certain functionalities like relaibility needs to be implemented at the endpoints to be correct and complete for this application

Question 40

Why can't the lower layers guarantee reliable delivery across the network?

Answer 40

They can't account for failures at intermediate nodes between sender and receiver.

Question 41

Why might implementing reliability at the core of the network be avoided?

Answer 41

Keeps core simple and cheap Reduces overhead for apps that dont need reliability Makes network more adaptable