Max NB-IOT Throughput

There are several aspects to take into account to obtain the effective NB-IoT maximum throughput for a single UE

UE category

The UE category which will limit the maximum transport block size. For NB1 UE, the maximum TBS for UL is 1000 bits in UL, 680 bits in DL. For NB2 UE, the maximum TBS is 2536 bits in both direction.

The particularity of NB-IoT is that allocations span several subframes. With the best modulation and coding and 12 subcarriers allocation, the maximum TBS is obtained in 4 subframes for NB1 in UL, 3 subframes in NB1 DL and 10 subframes for NB2 in both directions.

Scheduling

NB-IoT scheduling is by design full of 'gaps' that makes the effective throughput much smaller than the instantaneous bitrate of the NPDSCH/NPUSCH allocation.

For instance, in DL :

• There are 5 DL subframes min between NPDCCH and NPDSCH
• There are min 13 subframes between NPDSCH and NPUSCH format 2 HARQ feedback
• Plus 2 subframe minimum between NPUSCH and NPDCCH monitoring.
• Only 8/10 subframes are DL subframes (due to NPSS/NSSS/NPBCH)

So, for the 10 subframes 2536 bits NPDSCH allocation, this gives an estimated amount of(1+5+10)*10/8 + 13 + 2 = 35 subframes to transmit, so around 72 kbps throughput

This estimation is even optimistic since constraints due to SIB scheduling and MPDCCH periodicity will both reduce the subframes available for transmissions.

Scheduling improvements

3GPP have defined two major improvements for NB-IoT throughput :

• Non-anchor carriers (rel.13) can be used to maximize the number of DL subframes, since there are no more NPSS/NSSS/NPBCH or SIBs. This also allow for a much more regular scheduling
• Two HARQ process (rel.14) reduces greatly the 'gaps' in scheduling. For instance, in DL, a second NPDCCH can be scheduled between the NPDCCH and the NPDSCH to perform a second NPDSCH back to back with the first one. The penalty of the 5 subframes between NPDCCH/NPDSCH and the 13 subframes between NPDSCH/NPUSCH F2 is thus greatly reduced.

Amarisoft eNB performance (from version 2022-02-23)

As a summary, here is the effective throughput performance for Amarisoft eNB for each direction and for bidirectional trafic :

With a single HARQ process :

• 23 kbps DL for NB1
• 65 kbps DL for NB2
• 60 kbps UL for NB2
• 30 kbps DL / 30 kbps UL for NB2

With two HARQ processes on anchor carrier (NB2 is prerequisite) :

• 110 kbps UL
• 70 kbps UL / 50 kbps DL
• 85 kbps DL

With two HARQ processes on non-anchor carrier :

• 150 kbps UL
• 70 kbps UL / 70 kbps DL
• 110 kbps DL