Proportional Scheduling vs Legacy Scheduling

I’m trying to still wrap my head around the proportional scheduler vs legacy scheduler…

Speaking below of during a time of congestion, since scheduling doesn’t really care when the frames partially empty…

Legacy Scheduler basically takes all active data requests, and splits them evenly over the frame, basically making it so that lower modulation users get the same allocation as high users, resulting in low modulation users getting less bandwidth than high modulation users…

Proportional, tries to provide the same bandwidth to all clients, and adjusts allocations to fit that bandwidth so that say everyone gets 30% of their plan in that frame…

To maintain the current status of penalizing poor modulations, and maximizing throughput we would set the system to be locked BELOW THRESHOLD 256-qam… to recreate the conditions from legacy.

I believe the above is correct but wondered…

What if we wanted to penalize poor modulations more than the even split from legacy, would locking modulation at 6x (not below threshold, set to locked), result in 256-qam and 128-qam customers actually getting over allocated bandwidth in the frame vs their 1x/2x customers on the same panel to a higher level than on legacy (resulting in an overall gain in throughput on the panel, at the expense of poor modulation customers throughput).

Trying to clarify that in locked setup, lowering the modulation, gives a shifting scale of priority to the higher modulation clients, so say setting it to 7x would mean 256qam would get a slight boost over legacy and slightly affect everyone below 7x, 3x would DRASTICALLY shift traffic toward the upper modulations, and 2x/1x customers would basically get almost nothing?

This is a great topic for @LuciaCambium

During congestion, the legacy scheduler tends to allocate the same air time to each SM, regardless of their plans, while the proportional scheduler honors the ratio of the plans even when the link is fully utilized.
There are three configuration options for the proportional scheduler (see below for more details), but at a high level with User Lock Modulation disabled the ratio of plans is always honored, regardless of degrading or improving modulations on individual links. With User Lock Modulation Enabled or Enabled below threshold, the ratio of the plans is honored as long as the SMs operate at the current selected modulation; SMs with degrading modulation will see only their throughput reduced, while all others are not affected. The idea here is not to set the modulation level the same to all SMs, but to set it to the expected modulation for each SM. For example, assume you have three SMs operating at 2x, 4x and 8x. If the modulation lock is set to 8x for all, the SMs at 2x and 4x will reach 1/4 and 1/2 of their plan respectively. It is better to configure the modulation to the actual expected modulations (2x, 4x and 8x) so if the modulations are stable, each SM achieves the right proportion of their plan. If the SM operating at 4x degrades to 2x, for example, this SM will see its throughput decreased by half, but the other two SMs will keep operating at the same throughput.

  • Disabled: the scheduler will allocate resources to meet the plan, or a reduced value proportional to the other plans in case of congestion, regardless of the modulation.
    If the modulation of one SM degrades, the resources allocated to meet this SM’s plan increase, affecting the overall sector capacity. The reduced capacity is divided among SMs proportionally to their plan, but because the capacity is reduced, all SMs are affected.
  • Enabled: when selected, the Locked Modulation drop-down menu becomes available. Values for this menu are from 1x to 8x.
    The scheduler guarantees an amount of resources equal to what is necessary to transfer the data corresponding to the plan using the Locked Modulation. If the modulation decreases, the resources do not change, which means that the throughput to this user is reduced and no longer proportional to the plans. However, the throughput of the other users is not affected by one SM’s modulation degrading. Essentially, the SM’s plan is scaled down proportionally to the reduced modulation.
  • Enabled below Threshold: when selected, the Threshold Modulation drop-down menu becomes available. Values for this menu are from 2x to 8x.
    The scheduler behaves as in the Disabled mode as long as the SM’s modulation is above the Threshold modulation. If the modulation goes below the Threshold modulation, then the scheduler behaves as in the Enabled mode, using the Threshold Modulation as Locked Modulation.
    In this case, if one SM’s modulation degrades, all SMs’ throughputs are affected as long as the degraded modulation is above a threshold. However, as the modulation goes below the threshold, the resources are no longer increased for that SM, effectively capping the effect to the other users.

Well, our goal modulation is 8X for all SMs and due to small cell sizes, there’s no reason we shouldn’t hit 8X, which is the case for most of my clients customers, except when of course an antenna has moved due to weather, or an obstruction etc. We’d rather this customer not affect the overall sector performance during congestion, and we’re working on ways to secure overall sector performance and put more priority on fixing subscribers that are the random subscribers with lower modulations.

If I was to set 8X as a locked modulation expectation is that for the frame to be 100% utilized, we should see 120mbps (SU-MIMO) on throughput because no one is allowed to be over allocated space in the outside of an 8X frame time allocation, but in that same frame if we have customers with 4X they’ll only be getting a smaller share of their plan because they are below the locked modulation.

I really still have issues visualizing what the end result is between the Legacy/Locked/WithThreshold…

I’m trying to visualize how customers below and above the threshold/locked value are affected, vs each other and legacy, considering they’re all on the same package to simplify things… but having issues understanding the implications

If the expected modulation is 8x for all SMs, it simplifies things. The locked modulation and the locked below threshold cases will be identical if the modulation is set to 8x.
If you are expecting 120 Mbps total, and for simplicity we assume there are 4 SMs, then each SM will get 30 Mbps at 8x.
Assume now one SM drops its modulation to 4x.
The legacy scheduler, if there is enough traffic for all SMs and they have the same priority, will schedule them round-robin in each frame. The three SMs at 8x will use one frame out of four each, with a throughput of 30 Mbps, and the SM at 4x will use the fourth frame with half the throughput, 15 Mbps.
In this case, the proportional scheduler with a locked modulation of 8x (or locked below threshold) will have the same outcome. The three SMs at 8x will continue operating at 30 Mbps and only the SM with lower modulation is affected by its modulation drop, at 15 Mbps.
The proportional scheduler without locked modulation, however, attempts to provide the same throughput to all SMs. The airtime allocated to the user at 4x is increased so that all SMs can achieve the same throughput. Scaling everything, each SM can now achieve 24 Mbps, but the SM at 4x uses twice as many symbols as the others. So in this case SMs with the same plan achieve the same throughput, while in the other case SMs that meet the expected modulation achieve a higher throughput than the SMs that do not meet the expected modulation.
Things are different if the SMs have different plans. Assume you have three SMs with plans 100 Mbps, 50 Mbps and 50 Mbps. The total exceeds the 120 Mbps sustained by the link, so the scheduler needs to split the resources.
The legacy scheduler schedules the three SMs in a round-robin way, and if they are all operating at 8x, they will each get 40 Mbps. In this case the scheduler does not honor the SM with the higher plan.
The proportional scheduler maintains the ratio of the plans, and gives 60 Mbps to the first and 30 Mbps each to the other two.
Assume one of these two drops its modulation to 4x. If the lock modulation is disabled, all throughputs are lower, but they maintain the right propotion. The first SM gets 48 Mbps, and the other two get 24 Mbps each.
If the modulation is locked, then only the SM at lower modulation is affected. The first two will keep their throughput at 60 and 30 Mbps, while the one at 4x will drop its modulation to 15 Mbps.