Private LTE planning from a single project record: the Australian licence route with every condition cited to the RALI or instrument it comes from, uplink and downlink budgets with the limiting direction named, cell range across every propagation model held, and co-siting against the services already on the structure.
Included in the 14-day Pro trial on every new account. No credit card.
The errors in a network plan usually live in the joins, not the calculations. A band chosen for its licensing that the radio design then ignores. A bandwidth the throughput figure assumed and the budget did not. A range computed from the flattering direction. So everything here derives from one project record, and the code that assembles it is deliberately thin, because every calculation underneath is already tested on its own.
The licensing half is the part no other tool has. Given a band, a location class and a deployment shape, it names the routes that RALI MS 34, MS 47, MS 50 and SM 26 actually describe, and reports the conditions from the Radiocommunications Licence Conditions (Fixed Licence) Determination 2025 and the 3.4 GHz Advisory Guidelines with the clause on each. Where no held document offers a route it says so, and it distinguishes "no document covers this band" from "the documents cover it but offer nothing for this deployment". It never offers a plausible looking route that does not exist: band 43 sits between the two point to multipoint blocks and gets no point to multipoint path, however similar it looks to band 42, which does.
The budget half computes both directions with the PRB aware uplink noise bandwidth, names the limiting one, and hands that direction's path loss to the range calculation. Range comes back as a spread across models rather than a single figure, because they disagree and hiding that converts genuine uncertainty into false confidence. Site counts use hexagonal packing at the pessimistic end of the spread.
The co-siting half answers the question that actually stops private LTE deployments: can this carrier join a tower that already carries land mobile. It classifies the governing mechanism before calculating anything, because ACLR and ACS are only defined at the first and second adjacent channel and most co-sited services are tens of megahertz away, where the absolute out-of-band limit governs instead. It charges a narrowband victim only the leakage landing in its own bandwidth. And where the register cannot supply an input the answer depends on, it refuses to assess rather than substituting a plausible default.
Licensing, budget, range, throughput and co-siting all derive from the same record, and a test asserts that the band and bandwidth reaching each engine are the same ones. The joins are where plans go wrong, so the joins are what is tested.
Routes named from RALI MS 34, MS 47, MS 50 and SM 26, with conditions from the Fixed Licence Determination 2025 and the 3.4 GHz Advisory Guidelines. Every condition carries the clause it comes from, so a licensing position can be handed to a client and defended.
"No held document covers this band" and "the documents cover it but offer nothing for this deployment" are different answers with different next steps, and collapsing them into a single negative loses the one piece of information that tells you what to do.
Band 43 sits between the two point to multipoint blocks and gets no such path, however much it resembles band 42, which does. An unsourced licensing claim does not prompt a question, it prompts an application.
Computed separately with the PRB aware uplink noise bandwidth, and the range is solved from the limiting direction rather than the better one.
TR 38.901 RMa, UMa and UMi alongside a validated Okumura Hata kernel, with the disagreement reported as a ratio. The TR 38.901 implementation agrees with Sionna to within a millionth of a decibel across 456 cases.
Co-channel, adjacent channel or far offset, each governed by a different quantity. Applying ACIR at an offset where it is not defined is the commonest error in co-siting work, and the regime is decided first so the method follows from it.
The ACMA register has no receiver selectivity, no noise figure and no coupling loss, and those three decide the answer. A victim missing any of them is reported as no determination, naming what is missing, rather than being given a typical value.
Nothing, except that it has been done and every claim is sourced. The value is that the routes come from held documents rather than from how a band is usually used, that each condition carries its clause so it can be checked, and that the tool refuses to offer a route no document describes. Band 43 is the example that matters: it looks like band 42 and gets no point to multipoint path, because MS 50 confines that path to frequency blocks band 43 sits between.
Because no held document establishes a licensing position for it. That is not a claim that the band is unavailable, and it is deliberately not a claim that it is available. An unsourced availability claim does not prompt a question, it prompts a licence application, which is the expensive way to find out it was wrong.
It covers spectrum and licensing, a per site link budget and range, best server, RSRP and SINR across the whole network on a grid, capacity against an offered traffic load, PCI and PRACH allocation, and the emission half of co-siting. What it does NOT do, and says so on its own summary page: terrain, which none of its propagation models sees and which the Coverage Predictor does; antenna patterns and sectorisation; a frequency plan across sites, since the grid is reuse 1; and scheduler, QoS, control channel and backhaul limits.
The emission path: what the LTE carrier radiates into each licensed receiver on the structure, and what that does to its noise floor. Receiver blocking, intermodulation and PIM, and aggregate EME are separate mechanisms with their own tools, and none of them is run here. The result says that rather than implying the co-siting question has been fully answered.
Because the ACMA register does not carry receiver selectivity, receiver noise figure, or the coupling loss between two antennas on a structure, and those three are what the answer is most sensitive to. A victim missing any of them is reported as no determination, naming what is missing, so you know what to obtain. Substituting a typical value would turn a guess into a finding.
Because the models genuinely disagree, sometimes by a factor approaching two in area, and both ends are defensible for the same cell. A design that survives the pessimistic end is supported by the evidence. One that only works under the most generous model is a decision waiting to be found out, and a single figure hides which of those you have.
Yes, with a schema version on the record. An older payload opens unchanged because every field added since has been optional. A project saved by a NEWER build is refused rather than opened, because an older build would silently drop the fields it does not understand, autosave that loss back, and quietly diminish the project. A clean refusal is the better failure.
Included in the 14-day Pro trial on every new account. No credit card.
