Understanding Data Rate, MAC Protocols, and Network Technologies
¡Escribe tu texto aData rate =
Shannon: Max bitrate depends on signal/noise ratio
Nyquist: Max bitrate depends on n. sampling levels
CSMA CD
- Is a MAC protocol
- Exponential back-off strategy -> adapts to load on ethernet
- RTT bounds determines a bound on throughput
IP tunnels
- Packets traveling in a tunnel carry IP as load
- Allows for any IP to be sent to any destination (regardless of the IP carried as load)
- Never carries Ethernet as load
- Always carry IP packets as load
- The service offered by the IP protocol is unreliable.
Probability of error scenarios
- An error can occur if prob>0
- An error won’t always occur if prob<1
CAN APK
- Sender must always send a recessive symbol
- Sender is allowed to send a recessive
- Receiver is allowed to send a dominant symbol
- Receiver is allowed to send a recessive symbol
Switched Ethernet
- Switches carry out Ethernet frames forwardly
- The max. aggregate throughput of a switch equals the sum of the max. throughput of all input ports.
- Switches have no knowledge about IP addresses
- In case a switch is based on the concept of input-buffering, it can suffer from head-of-line blocking problems.
TTE (Time Triggered Ethernet)
- Multiple senders can send at the same time over the same network
- A rate-constrained frame can delay other frames in a switch.
- A best-effort frame can delay other frames in a switch.
- A time-triggered frame is never delayed by other frames in a switch. (If switches don’t shuffle.)
- A link is used half duplex.
MACA Ethernet
- MACA is a MAC protocol.
- The header of the RequestToSend frame carries information about the size of the data frame to send.
- If a host receives a ClearToSend frame, it will remain quiet until the corresponding data frame has been received by the destination host.
- In the MACA protocol, collisions can happen.
CAN, Flexray and Internet
- NOT possible to resolve contention on the Internet with CSMA/CR technique as used in CAN.
- The CSMA/CR protocol of CAN CAN’T also be used on a wireless network.
- NOT possible to resolve contention on the Internet with a static or dynamic segment as is used in Flexray.
- Answer is: None of the statements is true.
CAN
- Identifier (priority) is required to compute the minimum bandwidth allocated to the sender of frames with that priority.
CAN BUS
- Latency of frame with higher priority.
- Bandwidth of frame with higher priority.
- If bandwidth = 1, best case scenario, 100%
TT CAN
- Uses bit stuffing.
- Uses CSMA/CR.
- When frame is allocated an arbitration window, priority is needed to calculate max. latency.
- When frame is allocated 1 or more arbitration windows, n⁰ windows is needed to calc. max. lat.
FlexRay (for single and active star topology)
- Static
- Min. bandwidth computed with n⁰ slots.
- Min. bandwidth of sender is independent of other senders.
- All nodes have same slot.
- Dynamic
- Min. bandwidth computed with minislot number.
- All nodes have same minislot.quí!
PAR (Positive Acknowledgement with Retransmission)
- Efficiency = for
- Throughput(b/s) =
- Time out < RTT: Service is unreliable.
- Time out > RTT: Service is reliable.
- Wire like network: Reliable
Wireless communication in vehicles
- The Time Slotted Channel Hopping (TSCH) mechanism provides better robustness against interference from WIFI.
- The quality of low-power wireless links is in general unreliable and time-varying.
- The beacon-enabled mode of IEEE 802.15.4 standard provides both contention-free and contention-based medium access.
- Wireless signals with high frequency suffer from attenuation.
Worst case latency
- Sender does have priority: A is better than E.
- Sender does NOT have priority: E is better than A.
Polynomial F(x) CRC method
- Every burst error of length n or less will be detected.
- Error won’t be detected.
- The error is left undetected.
- The transmitted code is of form where is the remainder when dividing by modulo 2.
- Every error of form for will be left undetected because and this is divisible by
- The transmitted code is of form
- All errors with degree n-1 or less will be detected.
- Burst error of length n won’t be detected.
- is a bad choice, would be better.
- Only errors that occur in the redundant bit will be detected.
- Errors that occur in the original bitstream won’t be detected.
Spanning tree
- Purpose is to eliminate forwarding cycles (loops.)
- Dynamically changes with the physical topology.
- Disadvantage: may introduce unnecessary performance bottlenecks.
- Calculate paths with lower bridge number.
Aethereal TDM (answers to exam (always same exercise))
- In the same time slot circuits 1 and 2 can exit router R2.
- In the same time slot circuits 2 and 4 can exit routers R2 and R1 respectively.
- In the same time slot circuits 1 and 3 can exit routers R1 and R2 respectively.
- It is possible for circuit 1 to have a higher bandwidth than circuit 2.
- It is possible for circuit 1 to have a higher bandwidth than circuit 4.
Aethereal Network (components that have to be configured)
- Master bus (demultiplexor).
- Network interface and router.
Jacobson/Karels protocol
- Purpose is to determine dynamically the time-out time of the TCP.
- Computes estimations of expected values and standard deviations of segment RTT.
- and are used to tune the speed with which the algorithm will adapt to varying RTT.
Dijkstra algorithm (in this case for router F)
- Just calculate every path asked.
- For each X (not equal to F), the shortest path between F and X is unique.
Bit Stuffing
- 4B/5B encoding does not allow more than three consecutive 0’s, and provides higher channel efficiency than the Manchester coding scheme.
- TTCAN uses bit stuffing.
