In cellular technology.
The question always is and always will be about how much throughput a technology can achieve.
In order to answer, we need to specify the technology.
This discussion is focused on LTE.
For LTE. In Downlink throughput is received through Physical Downlink shared channel (PDSCH) .
The throughput achieved by the PDSCH depends upon:
The number of PDSCH Resource Elements depends upon:
Higher data rate depends upon modulation scheme.
Modulation scheme and quantity of redundancy depend upon the RF channel conditions.
UEs experiencing good channel conditions are more likely to be allocated higher order modulation
schemes with less redundancy
Multiple antenna transmission schemes (MIMO) increase the
throughput achieved by PDSCH.
2×2 MIMO using antenna port 0 and 1 approximately doubles the peak
throughput whereas 4×4 MIMO using antenna ports 0,1,2 and 3 approximately quadruples the peak throughput.
Cell specific reference signal overheads increase when using MIMO with
antenna ports 0,1,2 and 3 so the throughput are less than double and quadruple than the single antenna case.
PDSCH is a shared channel so its throughput capability is shared between all users.
Increasing the number of users reduces the throughput per user.
Users experiencing poor channel conditions reduce the overall total cell throughput .
Therefore as performance engineers and RF engineers you need to make sure users
experience minimal poor channel conditions.
The table below represents a set of theoretical absolute maximum physical layer throughput
which can be achieved if all Resource Elements are allocated to PDSCH channel and
the physical layer do not add any redundancy. ( This is not possible in practice , however
the table is provided as a reference from which you can have an understanding about
deriving the maximum expected throughput.
Table A: Non-Realistic Throughput Values ( Absolute Maximum FDD physical layer throughputs if all Resource Elements were allocated to PDSCH) Courtesy : LTE in Bullets
For non-MIMO throughputs values shown in the table A above.
These values have been generated by multiplying the modulation
symbol rate by the number of bits per symbol.
Looking at the bullet points for 20 MHz bandwidth above.
The Modulation symbol rate = 1200 * 14 / 0.001 = 16 Msps
The bit rate rate when using 64 QAM modulation scheme.
As there are 6 bits per symbol in case of 64 QAM modulation scheme
Therefore, 16.8 Msps * 6 bits per symbol = 100.8 Mbps
The MIMO throughputs in the table above have been generated by multiplying
the 64 QAM througputs by the relevant MIMO rank, i.e.
the throughputs have been double for 2×2 MIMO and quadrupled for 4×4 MIMO
In order to compute the maximum expected throughput is to remove the overheads generated by :
The table below shows the maximum physical layer throughput with the overheads removed.
The results are computed with a coding rate of 1. Which means the physical layer has
not introduced any redundancy.
The table B above illustrates the significant impact of the number of OFDMA symbols allocated to PDCCH, PCFICH and PHICH.
These physical channels can be allocated
In industry practice, the number of OFDM symbols allocated to PCFICH, PDCCH and PHICH depends upon the quantity of traffic loading the cell.
The values in table B are significantly less than table A.
For example, the throughput associated with the following channel
20 MHz channel + Normal Cyclic Prefix + 4×4 MIMO
decreases as follows
403 Mbps to 325 Mbps, 306 Mbps or 277 Mbps ( It all depends upon the number of symbols allocated to PDCCH, PCFICH and PHICH )
This means the impact of overhead generated by physical channels and physicals
signals does affect the throughput and more overhead reduces the application throughput.
As overheads do not transfer any application data.
The redundancy and overhead added to the physical layer reduces the throughput
measured at the top of the physical layer.
The PDSCH uses a combination of rate 1/3 Turbo coding and
rate matching to generated redundancy.
When UE experience poor channel conditions
When UE experience good channel conditions
The figure below shows how physical layer define the physical layer coding as a
function of channel conditions and modulation scheme.
Coding rate reflects the quantity of redundancy added by the physical layer.
Physical Layer coding rate as a function of channel condition and modulation scheme Courtesy : LTE in Bullets
QPSK and a Low coding rate are associated with poor channel conditions.
Link adaption allocates larger transport block sizes as the channel conditions improve ,
but the modulation scheme is kept as QPSK.
On account of this, quantity of redundancy is decreased and coding rate is increased.
Therefore large quantities of data are transferred
without increasing the capacity of the physical channel.
In the above picture of coding rate as a function of channel condition and modulation scheme.
Some things are visible
Once 64 QAM is allocated , link adaptation continues to allocate larger transport
block sizes as the channel conditions improve.
After 64 QAM there is no option to switch to a higher modulation scheme
once the coding rate reaches 0.75. However, link adaptation
continues to allocate larger transport block sizes and the coding rate approaches 1.
Some additional information not associated with application data is also transferred on PDSCH.
This information includes:
The overhead generated by SIB, paging messages and RRC signaling depends
upon the quantity of traffic loading the cell but is likely to be relatively
small i.e, less than 100 kbps.
This additional information reduces the capacity of PDSCH available for application data.
Retransmissions reduce the higher layer throughputs.
Hybrid Automatic Repeat Request ( HARQ ) retransmissions from the MAC layer
reduce the throughputs measured from above the MAC layer. A
utomatic Repeat Request ( ARQ ) retransmissions from the RLC layer
reduce the throughputs measured from above the RLC layer.
Similarly TCP retransmissions reduce the throughput measured from above the TCP layer.
Protocol stack headers also reduce the higher layer throughputs.
The MAC , RLC , PDCP and IP layers add headers to the application data.
The PDCP layer provides header compression for IP data streams so is able
to reduce the impact of the IP header.
The TCP and UDP layers also add their own headers when using TCP or UDP applications
In conclusion, the maximum throughput achieved on PDSCH is not a straight forward computation.
It all depends upon the scenario and assumptions for which you are computing the throughput available.
Also you read that based on the traffic in each Transmission Time Interval , the number of OFDM symbols can vary.
This can affect the overall throughput.
Therefore, in order to compute throughput on PDSCH in case of LTE FDD scenario. The most important point is to consider the context
in which you are computing the throughput and how much is the overhead in addition to actual application throughput.
In case if you are familiar with a scenario, please share it in comments below.
Session expired
Please log in again. The login page will open in a new tab. After logging in you can close it and return to this page.
information is simple and clear
Thank you, Benson. According to Ockham’s Razor theory its always good to keep it simple 🙂