Can any one explain difference between PDCP/RLC/MAC and Physical Throughput?
Which is has highest-rated value and in RF which is mostly consider and why?
Generally from drive test tool I remember we use to monitor PDCP DL throughput.
It is indeed a good question…
In my view PDCP is high as we move to RLC, MAC, PHY overhead, i.e. header added with data.
I mean at UE end, it is like user throughput.
Cell throughput from eNB.
PDCP, RLC, PHY - that is order of increased throughput.
Most useful throughput from customer perspective is PDCP.
PDCP is the lowest throughput.
Then RLC throughput is higher than PDCP throughput not only due to headers but also due to RLC retransmissions.
Then MAC throughput is higher than RLC throughput due to harq retransmissions and headers.
Then PHY throughput is higher due to headers.
From UE end or enode end this definition.
I think question is from DT tool throughput measurement.
It applies also to DT tools.
Ok, you mean to user throughput as well, i.e. Device measured throughput.
User throughput < IP throughput < PDCP throughput < RLC throughput < MAC throughput < PHY throughput
User throughput, what we get from kpi, is ipvolumeqci / iptime.
Cell throughtput vs User throughtput
As you know, throughtput is Data Volume divided by Transmission time.
If you consider Cell level, you divide data volume of the cell by the transmission time of cell: that means, the time where there is any data to be transmitted by that cell.
Now from user side, to calculate the User throughtput, you need to divide the user data volume by the time that UE is in transmission.
And to get the average user in a cell you need to divide the Sum of all UE data by the sum of transmission time of all UE.
Note that the volume data of all UE is the same Cell volume data.
If a cell has only one UE then Cell throughtput will be the same as User throughtput.
So what differs is the transmission time.
Any way to calculate user throughtput at PDCP, RLC, MAC and PHY level using counters in LTE network?
Guys, besides what you stated above, remember that User Throughput calculation, as stated by 3GPP, has some details regarding last TTI, which is discarded.
This is what we usually see in vendor KPIs.
Yes agree different vendors having different approach.
Huawei always win race in throughtput.
It’s all depend on algorithm and formula used.
One example, Ericsson using iptime combined by scheduling user fast without any gap so effectively throughtput inversely proportional to time. Less time more throughtput. One of area i seen how vendors playing with it.
Time = tti used for scheduling.
In theory, the formulas should be the one from 3GPP, but yeah, lots of room for optimization on that field
RF PDSCH throughput is widely used one in RF benchmarking purposes and basically, APP layer throughput is used for User perceived throughput from Session setup start to Session End.
And moreover since in 5G we have data divergence at PDCP for ENDC, we will get the performance of each radio interface for 5G & 4G from Physical throughput (PDSCH/PUSCH throughput) and due to header additions the tpt will be App layer < PDCP < RLC < MAC < PHY.
App layer = ip throughput. Is it correct?
For 5G part please tell are we monitoring PDCP or RLC layer throughput.
As per understanding in 5G PDCP layer give combined throughput for 5G and 4G?
Yes, App layer is IP layer, we take the HTTPTTx/Rx bytes/s for measuring APP layer throughput, and we preferred to use PDSCH throughput, even though RLC-MAC can also be used.
And PDCP thoughput was used earlier but mix of 5G PDCP alone and 4G PDCP alone cases came which compelled us to use PDSCH later.
What’s the value we are getting for throughput at PDCP level?
Which band combo used?
4G end cell user throughput.
5G end cell user throughput.
For ENDC case I am looking for value and which layer used for monitoring in network.
Need to clear if need to monitor throughput for 4G and 5G leg seperately.
Then whats approach used for ENDC.