ALERT TDMA Designer

What it is

A design tool for ALERT2 flood-warning radio networks. Give it the stations and the candidate sites and it works out where the receivers go, which sites relay to which, whether every gauge is actually heard at a base station in a design storm, and which time slot each transmitter should be given — then hands back a slot plan in the units an encoder is configured in. It runs as a hosted demo or an on-premises install, and it re-runs the schedule in milliseconds when a frame length, bit rate or bucket size changes.

TDMA slot design

ALERT2 stations transmit in assigned time slots against a GPS-disciplined frame. Every station has to be given a slot before it is commissioned, the allocation is static once it is in the encoders, and a shared channel in a flood is the busiest hour the network will ever see. The designer builds that allocation from the propagation model rather than from a spreadsheet, and says what it assumed to do so.

• Guard time is sized by clock drift during GPS holdover, not by distance. At 40 km the propagation spread is 0.13 ms; an hour off GPS at 1 ppm is 3.6 ms. The design states how long a station may run without a fix before it must stop transmitting.
• Slots are reused across space. Two stations conflict only if a receiver one of them relies on hears both — and a much weaker co-slot interferer is tolerated there. The conservative rule is computed alongside, so what the sharing bought is always a number.
• Repeaters are transmitters. Each holds slots for its whole subtree's storm rate, cannot receive while it transmits, and is placed later in the frame than everything it carries — so a relayed report never waits an extra frame.
• Capacity is dimensioned for the design storm, and compared against unslotted ALOHA, so the case for scheduling is shown rather than asserted.
• A legacy ALERT1 population on the same channel gets a contention window carved out of the frame, because a station that transmits on every bucket tip will transmit through anyone's slot.
• Re-schedule in about ten milliseconds: change the frame, the bit rate, the bucket, the sharing rule or the holdover, preview the difference against the saved plan, then save.

ALERT2 radio planning

Coverage in a flood-warning network is inbound: the question is never how far a base station reaches, it is from where a 5 W field station on a six-metre mast can be heard. The designer models exactly that, at 150 MHz, over real terrain, for every receiving site — and then follows each station's report through every repeater hop to a base station before it calls the station covered.

• Coverage is inbound and exact by reciprocity: the engine runs from the receiving site with the field station's power substituted, so the raster reads as power at the receiver.
• One raster, two readings. A design threshold with location, interference and ageing margins drives site count; an expected threshold is what a receiver would actually measure. They are never averaged and never confused.
• External noise is part of the budget per site — up to ten decibels between a quiet rural ridge and an urban roof — and the optimiser scores margin, never raw signal level.
• Verdicts follow the chain. A gauge is covered only if every hop to a base station works, and the report names the limiting leg.
• Trunk hops are held fifteen decibels higher than leaf links, because a failed leaf costs one gauge and a failed trunk costs a catchment.
• Solar is sized on the event, not the annual average: the gauge transmits most when the panel produces least.

What it produced on real catchments

Predicted results from the designer's own runs over Australian terrain. Catchments are anonymised; every figure below is a model output and not a measurement from an installed network.

The last row is the point of the tool. The spreadsheet answer for that catchment was "it fits". Counting the repeaters' own transmissions, it does not — and no frame length, bit rate or bucket size fixes it. The answer is fewer relayed reports, which is a siting decision, and the designer says so before anything is bought.

What you get

• A single map of every network with coverage, stations coloured by chain verdict, bases, repeaters and relay links.
• Per network: overview, TDMA design, capacity, relay proposal, power sizing and files.
• Slot plan CSV — one row per transmitter: serving and backup receiver, frame, slot length, guard, slot indices and start offsets, storm rate, holdover limit — every row marked as predicted.
• Survey report and Google Earth KMZ, station verdicts as CSV and JSON.
• Batch planning of an entire register: one plan per catchment, resumable, with a national roll-up.

What it is not

It is a prediction tool. Coverage, verdicts and slot plans come from a propagation model over public terrain and a register of station locations. No installed unit has been measured against it, the capture ratio and oscillator drift it assumes are engineering defaults it names on every page, and relay topology is proposed from measured paths rather than read from any record. Read its output as the starting table for commissioning, not as a survey of a working network.

Contact

Ask for a demo, a quote, or a plan for a specific catchment. Enquiries go straight to us.