Data Throughput - Ideal (Expected) Throughput

 

 

 

What is Ideal/Expected throughput ?

 

One of the most common questions that I have been getting for the past 10 years or so (probably from early HSDPA period and all the way up to now) was 'I have a UE with such and such category, what kind of throughput I need to get from this device ?' (Usually they carry the document like 36.306 with the tables showing each UE category and throughput).

 

One very important thing about the Category Table in 36.306 (LTE) or 25.306 (HSPA) is to understand that there are several assumption behind the table which are not explicitely documented. Followings are some of those assumption. (Actually the category table is for Phy/MAC layer standards, so iv)~vii) would not strictly relavent.. but you need to consider those items if you want to achieve good throughput at IP layer)

  • i) eNB or NB allocated all of its resource (meaning Maximum possible TBS for each TTI) for a specific UE (DUT)
  • ii) eNB or NB grants data transmission at every TTI
  • iii)There is no HARQ NACK nor DTX between UE and eNB/NB for each TTI (meaning no data retransmission)
  • iv) All the transport blocks are full of user data (No zero padding)
  • v) There is no NACK/Retransmission at RLC
  • vi) There is no Retransmission at IP layer
  • vii) RTT between UE and eNB/NB is minimum

If all of these assumptions are met, based on my experience you may expect to achieve at least 90% of throughptu at IP layer (UDP or TCP) comparing to the numbers in the 3GPP Category Table. If your UE and eNB/NB are well optimized, you may expect to achieve 95% or even higher.. but the amount of the effort to push from 90% to 95% would be huge.

 

Whenever you are doing the throughput test for your initial device, I always recommend you to check first if these assumptions are met before you are expecting any good throughput. At least, you have to check the first three item (PHY/MAC condition) for sure. It is no point of even thinking about iv)-vii) before you confirm on i)-iii).