DirectLinkNetworks
10 important questions on DirectLinkNetworks
What is needed for a physical connection?
Options
- point to point links
- shared access networks
Needed
- encoding bits
- framing
- error detection
- reliable transmission
- media access control
Questions
- what belongs in hardware?
- what belongs in software?
Explain how a node works
- encodes bits into signals, decodes signals into bits
- recognizes frames
- direct memory access (DMA)
- error detection
- medium access control (MAC)
Device driver: manages adaptor
Adaptor has a:
- input and output buffer
- control & status register (CSR)
Doest do: Reliable Transmission --> software
What are links, and what types are there?
- through a medium (twisted pair, coax, fiber) or through free space (wireless)
- wavelength = c / frequency
Local cabling
- Twisted pair (10-10000 mbps, 100m)
- Coax (10mbps, 200-500m)
- Fiber (100-100000mbps, 2-40km)
Carrier backbone & leased lines
- Copper based (1.544-44.700 Mbps)
- Optical carrier (50000Mbps - 200 Gbps)
- Wavelength-Division Multiplexing (WDM)
The Last Mile
- Twisted Pair
- Coax
- Fiber
- Wireless
- Higher grades + faster learning
- Never study anything twice
- 100% sure, 100% understanding
Difference ADSL and VDSL?
Since the evolution to VDSL, very high speed. Close to home there is a network unit. Twisted pair only goes there and the rest is fiber. The closer the unit to home, the bigger the bandwidth. FTTX: Fiber to the Neigborhood/Curb/Home
What is the internet checksum algorith?
Evaluation
- little redundancy (16 bits)
- easy to implement in software (but realistic for hardware?)
- compensating errors (+-) go undetected
What will CRC detect?
- All double bit errors: as long as C(x) contains a factor with at least 3 terms
- Any odd number of errors: as long as C(x) contains the factor (x+1)
- Any 'burst' error for which the burst length is less than k bits
- most burst errors of larger than k bits
Remark
- trivial CRC-1 reduces to even parity check
Options are; CRC 8, 10, 12, 16, CCITT, 32
How to ensure reliable transmission?
- assume: overhead of forward error correction (FEC) is too high
- corrupt frame must be discared
A reliable link-level protocol must be able to recover from discarded frames
Two fundamental mechanisms:
- Acknowledgement (ACK) of successful delivery
- Timeout to trigger retransmission of original frame
Automatic Repeat Request (ARQ)
- stop and wait
- sliding window
So THREE options
1: unreliable: frame is discarded and no correction
2: reliable: frame discarded and receiver gives no confirmation, so sender sends again
3: frame isn't discarded, receivers tries to correct it. With FEC.
What is acknowledgement & timeout?
-sent frame is received, send confirmation back
- sent frame is wrong and discarded. Sender waits for conformation that doesn't come. After timer it resends
Special
- confirmation gets lots, and resends frame unnecessarily. Receiver needs to see its double
- Everything works smooth, but timer is off, so it resends all the time.
What is Sequence Number Space?
- Sequence number space must be larger than number of outstanding frames
- SWS <= MaxSeqNum -1, is not sufficient if RWS = SWS
- SWS < (Maxsegnum + 1) / 2, is correct rule if RWS = SWS
Otherwise. It sends 7 things, 8 is the max but it hasn't gotten confirmation yet. This way it will endlessly resend.
What are some issues of Sliding Window?
Improvements
- Negative Acknowledgement (NACK)
- duplicate ACKs
- Selective Acknowledgement (SACK)
Sliding Window servers three different roles
- reliable delivery of frames over an unreliable link or network
- order preservation: higher level protocols receive data in correct order
- flow control: receiver controls speed of sender
Is it correct to incorporate multiple roles in a single protocol?
Sliding window is used in layer 2. Future we say layer 4 examples
The question on the page originate from the summary of the following study material:
- A unique study and practice tool
- Never study anything twice again
- Get the grades you hope for
- 100% sure, 100% understanding