Oaktown Tech / practical labs
Choose the split. Keep the service.
Compare 1:16, 1:32, 1:64 and cascaded splits. Trace the optical loss, keep the same subscribers, and find out what your next capacity decision really changes.
Read the text guide: instructions and limitations without interactive controls.
Experiment workbench
Return to the challenges ↑The starting values are invented for the exercise. They are not equipment specifications or recommended allowances.
Start with a question
Four identical PONs, 32 subscribers each, two equal active uplinks, and one 10 km optical path. Each button restores the baseline before making its one change.
Set the assumptions
Results update as you type. Busy-hour demand means an assumed average, not the speed printed on a subscriber’s bill.
Capacity assumptions 4 more inputs
Optical assumptions 10 more inputs
What the numbers say
↑ Change the assumptionsCompare a design decision
Keep one valid design in memory, change the assumptions, and compare the consequences.
Everything is calculated in this browser. Downloads are created here; no inputs are sent or saved. Import the JSON here later to restore these assumptions.
Make the planning decision
A smaller split buys room. Where will you put the extra fibers?
| Choice | What it can help | What to investigate |
|---|---|---|
| Centralized, single stage | Concentrates patching and makes reassignment between PON ports easier. | Distribution fiber count, cabinet space, access and feeder routes. |
| Cascaded, two stages | Shares distribution fibers out to smaller groups of premises. | Branch-specific spare outputs, extra field locations, joints and fault isolation. An empty output on one branch cannot serve a distant home without a physical route. |
| Lower total split | Less ideal splitting loss and fewer potential sharers per PON. | More populated PON ports for the same customers, optics, licenses, fibers and upgrade work. The aggregate traffic still reaches the uplinks. |
Take rate is not split ratio. If 32 homes are passed and 16 subscribe, take rate is 50%. Those 16 ONTs on a 1:32 splitter still see its full insertion loss. A 1:64 splitter serving the same homes does not make their take rate 25%; it changes output occupancy. Forecast connections by location, then repeat the capacity checks for busy hour and growth.
Follow one path. In a 1:4 × 1:8 tree, four second-stage splitters provide 32 outputs in total. Light reaching one ONT passes the first stage and one second stage. Losses add in dB; ratios multiply. Check the highest-loss path and the shortest path, in both directions, using the relevant wavelengths.
Do not divide the headline line rate by the split and promise that speed. A splitter divides optical power. PON traffic is scheduled on a shared channel. Usable throughput, bursts and service commitments need their own analysis. Unbalanced taps require a path-specific model and are outside this equal-split exercise.
References: Nokia: optical power budgets in PON and CommScope: centralized and cascaded architectures. All numeric presets in this lab are declared teaching assumptions, not specifications for either vendor.
Try to break one budget at a time
- Double the traffic. Which fails first: one PON, both uplinks together, or the surviving uplink? The fiber has not changed.
- Double the fiber. The traffic stays put. What happens to the optical reserve?
- Remove the spare. With one uplink, the failed-uplink case has no path left. No percentage makes that better.
- Test the other optical boundary. Restore the baseline, open Optical assumptions, and set minimum permitted loss to 24 dB. Why does adding reserve not fix the warning?
Show the worked answers
The baseline carries 1,920 Mbps down and 480 Mbps up per PON: 7,680 Mbps down and 1,920 Mbps up across four PONs. Its path loss is 23.3 dB; with 3 dB reserve, 1.7 dB remains against the entered maximum.
Double traffic exceeds the entered downstream capacity of one PON and of the surviving uplink. Both uplinks together still fit the arithmetic. Optical loss does not move.
Twenty kilometers gives 26.8 dB path loss and 29.8 dB including reserve: 1.8 dB over the entered maximum. Capacity does not move.
One uplink leaves none after its failure. Raising the minimum-loss input to 24 dB flags the baseline’s 23.3 dB component loss. A paper reserve does not add attenuation to the actual path.
What this lab deliberately does not prove
“Within” means only below the capacity you entered. It does not establish acceptable latency, bursts, queues, or customer experience. “At capacity” leaves no arithmetic headroom.
All PONs are treated alike. Uplinks are assumed equal, full-duplex, concurrently usable, and independent for the single-failure calculation. Per-flow hashing, shared conduits, device limits, protocol overhead, dynamic bandwidth allocation, and interoperability are not simulated. Aggregate bandwidth is not single-flow performance.
The optical check represents one path, one wavelength and direction, and one set of permitted limits. It does not derive limits from transmitter power, receiver sensitivity, or overload specifications. A real design needs the appropriate equipment specifications, demand measurements, construction records, and field acceptance tests.
Context: ITU-T XGS-PON reference. This exercise is not a certification of equipment, construction, security, or service performance.