Exam Prep Flashcards

0
Q

What are the disadvantages of simulation?

A

Often tempting to use simulation when not the best method of solving the problem
Doesn’t produce optimal results
It can be very costly

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
1
Q

What are the advantages of simulation?

A

Easily changed to see how modifications affect behaviour
May add to the understanding of the system
Simulation methods are easier to apply then analytic methods
In some cases the only way to study a system
Simulation data often less costly

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Give examples of objects, attributes and events?

A

Objects: Routers, Switches, Packets
Attributes: Processor speed, Packet size
Events: Packet arrives, Packet dropped

Interaction: attributes are linked to different objects and vary giving them realistic properties events occurs when objects experience change

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Challenges in multimedia traffic?

A
Compression 
Real time capabilities 
Inter-media sync 
Intra-media continuity 
Delays and jitters
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Elements,of a multimedia transmitter?

A
Source device 
Acquisition of data 
Data compression 
Inter media synchronisation 
Integration of streams
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Elements of multimedia receiver?

A

Media extractor
Decompression and reconstruction
Play back device

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the role of an extractor?

A

Received integrated streams and extracts into separate streams, eg audio and video

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Compare different types of jitter?

A

Intermedia: Time difference between packets of two streams eg A/V
E.g Audi video out of sync

Intramedia: Time difference between desired presentation and actual presentation of packets . Results in shaky picture or quivering voice because frames arrive early or late.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Contrast common compression techniques?

A

Header: only compresses packet header. Done hop by hop basis. Useful if pdus are smaller than headers.

Payload: whole body of packet. End to end.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Endogenous and Exogenous variables.

A

Endogenous: From within the boundary of the model. E.g, number of users, processor speed and bandwidth adopted.

Exogenous: from outside the boundary of the model. E.g, quantity of incoming data, power shortages.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Token bucket

A

Credits idle host
Token generated per tick of the clock
Preserves burstyness
Uses one token per object call

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Leaky bucket

A

Input varies output is constant
Doesn’t allow burstyness
Doesn’t credit idle host
Output is a smooth constant rate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Can a leaky bucket be used with a token bucket?

A

Yes! Leaky must go after token and leaky rate must be greater then that if the token bucket output rate.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

How is sequencing provided in MM?

A

RTP runs,on top of UDP to,provide time stamps and sequence numbers which help the receiver play back the packets in the correct order regardless of arrival time.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Why is TCP unsuitable for MM?

A

Never designed for MM or real time traffic
No QoS
Doesn’t provide critical timing
Overhead heavy
Aimed at apps with little sync necessity
No mutlicast

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Realtime transport protocol (RTP)?

A
Designed for real time traffic and MM
Runs on top of UDP 
Designed for mutli casting 
May be considered an application later protocol works between 4 and 7  
Audio and video need different streams 
Provides time stamping and sequence numbers 
Secure access via encryption
Payload identifiers
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Real time control protocol (RTCP)?

A

Works with RTP
Works between layer 4 and 7 of the OSI model
Sends reports on; packet loss, arrival, delay etc
Used mainly to modify sender rates and for diagnostics
Uses 5% of available BW, spits 75% to receiver and 25% to sender

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Congestion management concepts, policing and shaping?

A

Policing: propagates bursts, excess traffic is dropped - saw tooth pattern

Shaping: stores excess In a buffer,output is smooth. Relies on sufficient buffer size.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

QoS mechanisms (FIFO, PQ and WFQ)?

A

FIFO: excess packets stored in buffer and sent in order of arrival. No priority!

PQ: priority streams sent first (eg voice). Can starve low priory traffic and if all data is high priority, the high priority queue starts to discard.

WFQ: assigns streams a weight, with highest priory data getting a higher weight. Then uses round robin mechanism to take so many packets ( = to the assigned weight) from each stream as it goes.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

PCF wireless technique?

A

Uses Pifs
No contention : polls (pc)
Can be used with DCF in the same BSS

20
Q

How to overcome the WFQ buffer problem?

A

Move discarded packets into a lower priory queue
Alter weight of queues in adaptive fashion based on queue size
Allow dynamic buffer allocation

21
Q

Random early detection (RED)?

A

Monitors traffic loads
Randomly discards PDUs if congestion begins to increase
One type is weighted RED (WRED)

22
Q

What is weighted RED (WRED)?

A

Combined RED with IP precedence
Selectively discards lower priority traffic when interface starts to get contested
Dropped on criteria; queue depth, source etc

23
Q

Types of delay?

A

Process: time taken for router to receive PDU on out interface examine it and place it on in

Queue: time spent in output queue

Serialisation: time for bits to be put on huge wire

Propagation: time to travel across a link

24
Q

Temporal and spatial compression?

A

Temporal: removes redundancy between consecutive frames as they are almost the same
Spatial: done independently inside each frame by removing redundancy from the frame.

25
Q

802.11e?

A

Built on 802.11 MAC
Introduces new hybrid coordination function (HCF)

Two types:

EDCA: like DCF but introduces different priorities for services

HCCA: like PCF but with priority

26
Q

Enhanced distributed channel access (EDCA)

A
Divides traffic into different priority queues -access categories (AC)
Each queue gets different:
- minimum contention window 
- maximum contention window 
- arbitrary inter-frame space (AIFS)
- transmission opportunity (TXOP)
27
Q

Advantages of EDCA?

A

Voice and video have priority over data

Works well if network has light load

28
Q

Disadvantages of EDCA?

A

Lower priority traffic starvation
Streams of same priory contest
Not able to guarantee access based on latency, BW etc

29
Q

Distributed coordination function (DCF)?

A
Uses basic CSMA/CA mechanism 
Random back off - binary exponential 
Uses inter frame space:
-SIFS: shortest, used for CTS/RTS messages
-PIFS: medium, used for PCF
-DIFS: normal (longest), used for data
30
Q

What are virtual collisions?

A

If the backoff counters of two or more parallel ACs in one QSTA reach zero at the same time, a scheduler inside the QSTA will avoid the virtual collision by granting the EDCA-TXOP to the highest priority AC. At the same time, the other colliding ACs will enter a backoff process as if there is an external collision

31
Q

Difference between peak rate and burst size?

A

Peak rate: measured in B/Sec. Max data rate of traffic

Burst Size: measured in sec. Time traffic is generated at peak rate

32
Q

EDCA access categories?

A

In order of increasing priority:

Background traffic
Best effort
Video
Voice

33
Q

Which conditions allow a frame to be sent immediately?

A

If queue is empty and edit has been idle longer than the AIFS

34
Q

Calculation involving video?

A

Frame size X bit rate X frame rate

/BW of link in B/Sec

35
Q

Calculate the bandwidth needed and compression ratio

A

Frame size X bitrate X frame rate
/ compression ratio

Round up to allow excess

36
Q

What does RTSP provide?

A

Used in multimedia to provide extra Streaming functionalities like play and pause
Controls multimedia communication
Out of band

37
Q

Describe echo

A

Annoying effect you here on a phone call when you hear your own voice playing back after a period of time. Echo if a few milliseconds is bearable but a few 100 milliseconds disrupts the call severely.

For echo to be noticeable it has to be loud and delayed. Delay is present in PSTN but worse in converged VoIP networks as they are more prone to delay. PSTN has less than 10 milliseconds delay where as VoIP can have 400 milliseconds.

38
Q

Stages of methodological approach

A
Defining the problem
Modelling
Programming
Experimentation 
Implementation and monitoring
39
Q

Defining the problem

A
The problem must be defined
What is the object of the exercise?
What level of accuracy is required?
The grater he accurate the more data required and the higher the cost
What data must be collected?
What form should the solution take?
Is simulation the correct approach?
Considerations of safety is paramount
Carry out cost benefit analysis at regular intervals
40
Q

Modelling

A

Can involve compromises and decisions
Who will do the modelling?
What level of detail is needed?
If opnet is being used an in screen model can be created.
If alternative simulation used activity cycle diagram must be constructed.
Where possible data collected by modeller should be used.
The speed the model and results are to be provided to management?
Time scales must be dynamic as more info is available
Verification takes place here to ensure management accept model as appropriate and correct

41
Q

Programming

A

Programming involves the representation of the model, in a computer language.
Choice of language used decided at formulation stage but can be changed.
In general special purpose languages are used
Coded program has to be verified
Must be documented in detail
Validation occurs here, to ensure model is accurate representation of the system etc

42
Q

Experimentation

A

Define the base run of the model.
Run the media in a steady state and transient states and collect
Consider sensitivity of model
Discuss output with management

43
Q

Implementation and monitoring

A

Small amounts of info to be given regularly to those involved
Don’t try to please everyone

44
Q

802.11e HCF introduces two possible modes of operation

A

EDCA: like DIFS but with priority levels
HCCA: is a CSMA/CA polling based method like PCF but without the shortcomings of PCF

45
Q

Five principles of model building

A

Relevance: should only include relevant aspects if the system
Clarity: clear statement of objectives and assumptions
Accuracy: how accurate diss it have to be? How much data?
Available data: data must be readily available
Cost: limited funds, very important factor, weak model that is complete is better than an incomplete stronger model

46
Q

Iterative nature of the 5 phase methodology

A

Each phase on its own is iterative S questions arise which may require you to go back to the start if the phase
Iteration occurs between phases where progression may not be linear and several stages may be repeated
Worst aces scenario all stages have to be done again if solution is not appropriate.

47
Q

How is time stamping provided by UDP?

A

Each packet has a time stamp that shows the time it was produced relative to the first or last packet, the receiver can add this time to the playback time to separate arrival time from playback time. UDP has no time stamping but RTP runs on top to provide this