Adaptive vs Round-Up: Are You Sacrificing Throughput or Efficiency Without Realizing?

[J_K]

Hi experts - I’d like your advice on configuring the Adaptive RBG Allocation Strategy parameter.

From the documentation (Huawei):

• ROUND_DOWN → fully utilizes allocated RBs, improving spectral efficiency, but results in lower cell throughput
• ROUND_UP → leaves some RBs unused, reducing spectral efficiency, but improves cell throughput

Given this trade-off, what is the recommended setting in real network conditions?
Should this parameter be configured as Adaptive or Round-Up, and in which scenarios would each option be preferred?

I’m especially interested in practical deployment insights (e.g., congestion vs capacity-driven environments).

[rizwankader]

Out of curiosity, what’s making you consider changing this parameter?

In most cases, Adaptive works well as the default. But if your BLER is already quite low, it might be worth trying ROUND_UP to squeeze out a bit more throughput.

[J_K]

To improve overall network performance.

[rizwankader]

What exactly is the issue with the network? Is it accessibility, retainability, load, or throughput?

[J_K]

The main issue seems to be load and throughput.

[rizwankader]

How many carriers do you have?

Are all of them heavily loaded?

[J_K]

Almost all the main carriers are heavily loaded - the radio link is already reaching around 300 Mbit/s in terms of transport capacity.

[rizwankader]

What I meant is: are the RF carriers (bands) evenly distributed in terms of load, or are some bands more congested than others?

[J_K]

CA (Carrier Aggregation) is active.

[Geoff]

Has anyone tried using ROUND-UP together with PreciseRbAllocTbsIndexDec@CELLDLSCHEXTPARAM (recommended value: 6)?

I came across this combination in the documentation, but I’m curious about real-world behavior.

Does it actually improve throughput under load, or does it introduce any side effects (e.g., BLER impact, scheduler instability)?