fix: start window modification during qer rate limiting

[tariromukute]

Description:

This pull request addresses a critical issue in the UPF QoS implementation where the QER rate limiting was not functioning correctly. The qer->ul_start and qer->dl_start value, used to track the sliding window for rate limiting, was never updated, resulting in uncontrolled traffic rates. This PR corrects the rate limiting logic by properly updating the sliding window start time.

Problem:

The limit_rate_sliding_window function was incorrectly modifying the pointer to the start time, rather than the start time itself. This resulted in the start time remaining at its initial value (zero), effectively disabling rate limiting.

Solution:

This PR modifies the limit_rate_sliding_window function to directly update the windows_start value.

Testing:

To validate these changes, please refer to the testing instructions in the repository: https://gitlab.com/phine-oss/upf-qos-validation#251-integrating--eupf-ueransim-and-ues

Results:

• Before: The traffic rate was not capped and remained uncontrolled because qer->ul_start was always zero.
• After: The traffic rate is now capped according to the QER parameters.

Known Issue/Further Improvement:

During testing, it was observed that the traffic capping was not perfectly accurate. For example, a requested rate of 20 Mbps resulted in approximately 13 Mbps. This discrepancy requires further investigation and potential refinement of the rate limiting algorithm. The current fix ensures that the rate is limited, but the exact accuracy requires further improvement.

[pirog-spb]

Hi @tariromukute

Good PR! Thanks!

I'll create additional issue concerning inaccurate rate limiting. I guess, it's because of lost of accuracy during int - float - int conversions during rate limiting.

[tariromukute]

Hi @tariromukute

Good PR! Thanks!

I'll create additional issue concerning inaccurate rate limiting. I guess, it's because of lost of accuracy during int - float - int conversions during rate limiting.

Hi @pirog-spb. Thanks for the review and for creating the follow-up issue!