ch8 Flashcards
CDMA vs. FDMA vs. TDMA
Understand the differences
FDMA = Frequency Division Medium Access
2 * 25 MHz = 2 * 125 frequencies
User gets a specific frequency
\+ Continuous transmission
- Only use a single frequency ("narrowband")
TDMA is on top of FDMA
TDMA = Time Division Medium Access
8 channels
User gets fraction of the available capacity
For instance, user gets time slots 2, 10, 18. 26, etc.
+ User can use complete frequency band (“wideband”). Take in mind that this one of the frequency bands of FDMA
- No continuous transmission
Universal Mobile Telecommunication System (UMTS)
Uses FDMA with 2 * 60MHz = 2 * 12 frequencies
Uses CDMA (Code Division Medium Access) instead of TDMA:
Multiple connection can still use the same frequencies.
Devices all communicate at the same time, but using different “languages” IMPORTANT
More users, means more noise
+ high data rates + continuous transmission + easy network expansion
- Interference (More noise) and “near-far” problem. Power control needed. The further you are away, the more noise
Slide 9
Shared channel vs. scheduled transmissions
Understand the differences (between UMTS and HSDPA)
UMTS:
- Each flow is assigned a thin dedicated channel
- Parallel transmissions
HSDPA: (High-Speed Downlink Packet Access)
- Is an extension of UMTS
- All flows share a thick “shared channel”
- Transmissions are scheduled, rather than parallel
- Opportunistic scheduling
- Needs additional hardware and software
Slide 13
Opportunistic scheduling
Know what it is
This is used in HSDPA
An algorithm decides what device gets access to the medium (based on signal/noise-ratio)
So the device with the best signal/noise ration will be scheduled, since this would technically be best for the performance of the system.
By itself, this is not completely “fair”
High-Speed Download Packet Access (HSDPA)
Understand the concept of the technology
HSDPA: (High-Speed Downlink Packet Access) (3.5G)
- Is an extension of UMTS (or successor of)
- All flows share a thick “shared channel”
- Transmissions are scheduled, rather than parallel
- Opportunistic scheduling, by keeping track of the quality to each device
- Needs additional hardware and software. (Large investment necessary)
- Higher throughputs
- Mostly used in big cities (at the time, we now have 4G/5G). As you go to more rural areas, signal may switch to UMTS and after to GSM/GPRS
Misconceptions
Understand why advertised bandwidths are overly optimistic
Reporters/people who don’t fully understand the technology usually only look at the direct bitrate
But this bitrate is only possible when you are extremely specific conditions (such as standing next to the base station while no-one else is connected).
In reality, the rate is shared.
Propagation aspects (SINR, multi-path fading) Understand the concepts
The received signal depends on the distance to the base station.
This also depends on how many “objects” may block the signal. (eg. buildings)
Signal-to-interference-plus-noise ratio (SINR):
- Determines the assigned bit-rate
- Near base station, means high SINR
- Near cell edge, means high SINR (more noise from different cells and further away from the base station)
- Outside of range of base station, and closer to others (This will usually make you switch base station)
Multipath fading
- User may receive copies of the signal that are reflected of “objects”
- They may strengthen or weaken the signal
- Very difficult to model
Rate adaptation
Understand the concept
In HSDPA, when a flow is served. The Rate Adaptation scheme decides what bit rate the flow will get (at a 2ms time scale)
So it decides which terminal gets access to the medium, every 2 ms
The higher the signal/noise ratio, the higher the rate they get. (So depends on the average SINR, and as such the distance to the base station)
Single-cell ‘onion model’ for HSDPA (concentric circles)
Understand idea of the model
Since the SINR depends on the distance to the base station.
The rate you can is defined by the distance.
The rate you get is determined by what “layer” you are in, with the base station as the centre
Slide 29
Packet scheduling (RR, max SINR, proportional fairness)
Understand what the pros and cons are (fairness vs.
efficiency)
Packet scheduler decides which flow is served.
Round Robin (RR)
* Cyclic service of flows
+ Inherently fair, every device gets equal time
- Inefficient
Maximum SINR:
* Device with best signal/noise ratio is served
- Unfair, distant flows may not be served, or served less
+ Efficient, because the best signal gets more time
Proportional fairness (PF):
* Always serve flow with best RELATIVE SINR (compared to average)
+ Efficient, somewhat less then Max SINR
* Medium fairness
Multi-user diversity gain
Know what it is, and understand the idea
How much more efficient do you get the total average data rate compared to the naive way.
“Multi-user diversity gain uses the fact that there’s multiple users and this diversity in terms of signal is boosting the capacity of the highest” - Rob (Whatever that means)
Didn’t really understand it
How much you do better in terms of throughput in comparison to Round Robin
M/G/1 multi-class Processor Sharing model Know what it is, know that is has a product-form solution and understand why it nicely models the performance of HSDPA
Effective capacity depends on SINR
We have defined circles dependent on distance to base station.
Probability of being in a specific circle, is given by the size of the circle.
- So arrival rate in a specific circle is calculated using lambda and this probability
Per zone there is multi-class processor sharing. All the calls within a circle share the capacity equally (I think, perhaps check this to make sure!)
It has a product-form solution :)
