The 900 MHz Band Replan: What Australia's Decision Means for IoT, Mining Radios, and Smart Meter Networks

The short version

Australia’s 900 MHz band is being rebuilt. ACMA has reorganised the 820 to 960 MHz range to align the cellular paired bands with the global 3GPP plan, harmonise duplex spacing, and free up contiguous spectrum that was previously occupied by a patchwork of incumbents: 2G GSM, point to point STL links, fixed and land mobile narrowband, RFID, and a long tail of low power class licensed devices.

The end state is a cleaner cellular allocation, a re purposed lower edge for railway and mission critical broadband, and a redefined class licence regime in 915 to 928 MHz.

For an RF engineer working in Australia, the practical consequences land in three places:

• Paired band cellular operators have to retune and recoordinate.
• Incumbent narrowband apparatus licensees are being moved or sunset.
• Class licensed IoT in 915 to 928 MHz now operates under tighter rules, with emission masks and duty cycle limits that change link budgets.

This article walks through each in detail.

---

Status and sourcing: read this first

This article is an engineering interpretation, not a regulatory summary. The 900 MHz replan spans several ACMA processes that have run over multiple years, and the underlying documents sit at different stages of finality. To keep the engineering discussion readable, the band edges, sub block boundaries, EIRP ceilings, emission masks and listen before talk parameters below are presented as a single coherent picture. In reality they come from different instruments at different stages.

Treat the specifics in this article under four labels:

• Finalised allocation. The broad cellular re banding toward the 3GPP Band 5 and Band 8 plan, and the lower edge public safety and rail allocation, are settled policy direction.
• Consultation proposal. The detailed LIPD sub block structure, EIRP steps and LBT thresholds reflect the direction of ACMA consultation rather than a clause you can cite from an in force instrument today.
• Likely implementation outcome. Migration timetables, and the relocation of STL and fixed links to 7 and 8 GHz, are the probable path based on how ACMA has handled comparable re bands.
• Engineering assumption. Isolation figures, selectivity numbers, coordination distances and link budget arithmetic are our own synthesis. They are worked examples, not regulation.

Before you commit capex or file a coordination submission, verify every parameter against the primary sources:

• the Radiocommunications (Low Interference Potential Devices) Class Licence, the in force instrument and any amendment, on the Federal Register of Legislation and acma.gov.au
• ACMA consultation papers and outcomes for the 900 MHz band re band and the LIPD review
• the relevant RALI (Radiocommunications Assignment and Licensing Instructions), in particular the MS series for fixed and land mobile coordination
• the Register of Radiocommunications Licences (RRL) for your own assignments
• ACMA’s current five year spectrum outlook for timing

Where this article states a number, assume it needs that check.

---

The replanned band at a glance

The table below reads as a single frequency axis from 803 to 960 MHz, lowest segment first.

All cellular pairs use a 45 MHz duplex split. The LIPD class licence segment sits inside the n8 duplex gap: hard against the n8 uplink edge at 915 MHz, and 7 MHz below the n8 downlink edge at 935 MHz.

---

Where the band sat before

Before the replan, the Australian 900 MHz region was a layered allocation that had grown organically since the 1990s:

• 820 to 825 / 865 to 870 MHz: a mix of fixed and land mobile narrowband apparatus licences, some Public Safety Mobile Broadband trials, and historical defence assignments.
• 825 to 845 / 870 to 890 MHz: cellular Band 5 paired uplink and downlink, dominated by Telstra and Optus, with legacy 2G CDMA and GSM holdings transitioning to 3G then LTE.
• 890 to 915 / 935 to 960 MHz: GSM 900, P GSM and E GSM bands, again largely held by the major carriers, with several megahertz at the edges used for STL (studio to transmitter link) and fixed point to point services under apparatus licence.
• 915 to 928 MHz: the LIPD (Low Interference Potential Devices) class licence segment. This is the Australian analogue to the US 902 to 928 MHz ISM band, but narrower by 13 MHz and with its own emission rules. It was the home of LoRa, Sigfox, Wi SUN, ZigBee variants, RFID readers, telemetry SCADA links, smart meters, and a long tail of consumer telemetry.

Two structural problems drove ACMA to act, and both were obvious to anyone who had tried to coordinate a new service in the band over the last five years.

First, the cellular paired allocations did not cleanly align with the 3GPP Band 8 plan used by every modern device. That made carrier aggregation and refarming inefficient and pushed handsets onto suboptimal RF front ends.

Second, 915 to 928 MHz had become densely populated with class licensed LPWAN gateways, RFID portals, and smart meter concentrators. The assumption of “low interference potential” had stopped holding in many urban and large industrial environments. Mining sites and metropolitan logistics yards in particular have reported noise floors elevated well above thermal in the upper half of the band, in some measurements by 8 to 15 dB.

---

The new band plan, layer by layer

The replanned 900 MHz region is best understood as four contiguous zones, each with a different regulatory regime.

Zone 1. 803 to 824 / 848 to 869 MHz: the new lower edge for public safety and rail

This block is now reserved for public safety mobile broadband and railway communications. The rail standard is FRMCS, the Future Railway Mobile Communication System, which replaces GSM R globally.

Existing land mobile apparatus licences in this range are being migrated out on a published timetable.

For an engineer designing a private rail or critical infrastructure network, this is the band that matters. The duplex split is 45 MHz. The channelisation is expected to support both narrowband 200 kHz carriers for legacy interoperability and 5, 10 or 15 MHz LTE and NR carriers for broadband payloads. This is also the most likely home for private LTE serving critical infrastructure operators who qualify under the eligibility rules.

Zone 2. 825 to 845 / 870 to 890 MHz: cellular Band 5 / n5

Unchanged in spectrum terms, but cleaned up administratively.

The legacy 2G assignments have been formally retired, and the paired allocations are now expressed in the 3GPP n5 channel raster. For RF planning this means existing Band 5 antennas, combiners, and TMA filters remain valid. Link budgets should still be re run against the emission masks in the adjacent zones above.

Zone 3. 890 to 915 / 935 to 960 MHz: cellular Band 8 / n8 expansion

This is the headline change for the carriers.

The full 25 MHz paired GSM 900 allocation is now harmonised with 3GPP Band 8, with carrier aggregation tested against Band 28 (700 MHz) and Band 3 (1800 MHz). Telstra, Optus, and TPG hold the bulk of this spectrum following the most recent allocation round.

STL and fixed point to point services that previously occupied the band edges are being relocated to alternative bands, commonly 7 GHz and 8 GHz for new licences, with a transition window understood to run through 2027.

If you operate an STL link with a renewal coming up in the next 18 months, this is the licence to look at now. Re engineering a studio to transmitter hop at a new frequency typically requires:

• a new path profile
• a new fade margin calculation, accounting for the higher rain attenuation at 7 or 8 GHz
• in many cases a new antenna with tighter beamwidth, to meet the coordination distances at the higher frequency

Zone 4. 915 to 928 MHz: the redefined LIPD segment

This is the layer that will affect the largest number of engineers, because it is where almost all class licensed IoT operates.

The sub block structure, EIRP ceilings and access rules below track ACMA’s consultation direction rather than an in force class licence.

The proposed LIPD class licence retains the 915 to 928 MHz allocation but introduces a sub block structure rather than a single uniform allocation:

Three structural changes matter for design:

Channelisation. The band is divided into the three sub blocks above, each with different power and access rules. Most LoRaWAN and Wi SUN gateways are expected to move to the 920 to 926 MHz sub block to use the higher EIRP ceiling.

Emission mask. Out of band emissions are proposed to sit at least 36 + 10 log P dBc below the in band power across the immediately adjacent 1 MHz, tightening from the previous flat 30 dBc rule. For a 1 W (30 dBm) transmitter this means out of band power below roughly 6 dBm in the first adjacent megahertz. Some cheaper LPWAN gateways with relaxed front end filtering are likely to struggle with this. Designs that comfortably passed the previous flat 30 dBc rule should be re measured rather than assumed compliant.

Listen before talk. Devices operating above 25 mW EIRP in the 920 to 926 MHz sub block are proposed to implement listen before talk (LBT), with a sensing window in the order of 5 ms and a threshold near 80 dBm/MHz. This is a meaningful change for LoRaWAN networks, which historically rely on pure ALOHA random access. Affected deployments would need to be reconfigured to use the LBT modes in the Semtech reference stacks.

The duty cycle and LBT changes have a direct effect on link budgets and capacity planning.

A LoRa SF12 uplink at 125 kHz occupies the channel for around 1.5 seconds. Under a 1 percent duty cycle in the lower sub block, that limits a device to roughly 24 transmissions per hour. Moving to the 920 to 926 MHz sub block lifts the EIRP ceiling to 4 W but mandates LBT. Achievable goodput then depends on local channel occupancy rather than a fixed duty cycle. In lightly loaded deployments the LBT regime is more generous. In noisy industrial environments it can be more punishing.

---

What incumbent apparatus licensees actually need to do

The replan is being implemented through a published migration framework. The headline transitions:

On the STL and fixed links, ACMA has indicated it does not intend to renew at the existing frequency past the transition window. Treat the exact cut off as something to confirm against the current RALI and any published migration notice rather than from this article.

Legacy 2G assignments are retired. Any remaining 2G IoT modules, including some older smart meter fleets, alarm panels and lift telemetry, need a replacement path. The realistic options are LTE Cat M1 or NB IoT on 700, 850 or 900 MHz, or a private LPWAN on the redefined LIPD allocation.

For an engineer holding an existing apparatus licence in this region, the immediate action is to pull the current assignment from the RRL and check three things:

1. the licence expiry
2. the licence type (fixed, land mobile, STL)
3. whether it falls inside one of the migration zones

If it does, the cost of moving needs to be in the next capex cycle, not the one after.

---

Coexistence reality: LoRaWAN, NB IoT, LTE M, Sigfox and Wi SUN

The redefined LIPD segment sits immediately adjacent to the cellular paired band.

The downlink edge of n8 is at 935 MHz, which is 7 MHz above the upper edge of the LIPD allocation. In practice, LPWAN gateways and cellular base stations end up colocated on the same rooftops, towers and shelters. The in band to in band interference question becomes a real engineering problem.

Two coexistence issues come up repeatedly.

LoRaWAN and Wi SUN gateway desensitisation from cellular downlink

Consider a high gain LPWAN receiver at 922 MHz. It is looking at an n8 downlink carrier at 935 MHz, 13 MHz away. On a shared structure, that receiver needs in the order of 70 dB of front end selectivity to avoid blocking.

Many off the shelf gateways provide closer to 50 to 55 dB. That gap is the kind of margin that decides whether a site works.

The practical fixes:

• physical separation, typically around 3 m vertical or 8 m horizontal at the same height
• an inline cavity filter on the LPWAN feed
• moving the gateway to the upper sub block, where the offset to the cellular edge is larger

A worked blocking budget for a shared site

The qualitative argument above is easier to act on with numbers. Here is a simplified blocking budget for a LoRaWAN gateway colocated with a cellular n8 downlink on the same structure.

The isolation estimate and the gateway blocker tolerance dominate this result, so measure both on the actual hardware before trusting the margin.

With these inputs the gateway is desensed. Two ways to recover the 8 dB:

• Filter. A cavity filter on the LPWAN feed with around 30 dB rejection at 935 MHz drops the blocker to roughly -52 dBm, about 22 dB clear of the desense onset.
• Geometry. Lifting the vertical separation from 3 m to about 5 m adds close to 9 dB, since the vertical isolation relationship scales as 40 log of the spacing.

The cost of doing nothing is direct. An 8 dB desense removes about 8 dB of link budget. A concentrator with an SF12 sensitivity near -139 dBm then behaves like one rated near -131 dBm, which is a visible loss of range at the edge of the coverage area.

Cellular uplink desensitisation from LPWAN bursts

This is the less obvious direction, but the one that tends to bite in dense deployments.

A 4 W EIRP LoRaWAN gateway transmitting in 920 to 926 MHz produces broadband phase noise and intermodulation products. Some of those can fall in the n8 uplink at 890 to 915 MHz.

The 14 MHz guard between the LIPD upper edge and the n8 uplink upper edge looks comfortable on paper. But if the gateway runs a wideband chirp and the cellular receiver is at the edge of its sensitivity, the link budget can close uncomfortably.

Where NB IoT and LTE M sit

NB IoT and LTE M deployed on the carrier networks live inside the licensed cellular bands, so their coexistence with the LIPD segment is the carrier’s coordination problem, not yours. The exception is a private NB IoT or LTE M network in the new public safety and rail zone, where the same colocation arithmetic applies and you own it.

The coordination checklist

In both interference directions, the practical engineering controls are filter selectivity at the colocated equipment and explicit coordination distance rules on shared sites. For a noim3 frequency coordination workflow on a mixed cellular plus LPWAN site, the relevant checks are:

• isolation per port, typically 60 dB minimum for class A receivers
• TX noise floor at the victim band, below thermal plus 6 dB
• intermodulation products to the third order, across all carrier combinations

Treat these as starting points and tune them to your own equipment.

---

A worked example: SCADA telemetry for a regional water utility

A regional water utility runs 800 SCADA endpoints across a 60 km service area. The legacy system uses 2G GSM modems for hourly meter reads and 400 MHz UHF radios for pump telemetry.

With 2G retired and 400 MHz under coordination pressure, the engineering choices reduce to three.

1. LTE Cat M1 on the carrier 900 MHz network. Coverage is provider dependent. In this service area it would mean roughly 92 percent endpoint reach, with the remaining 8 percent needing yagis or an alternative bearer. Per endpoint cost is dominated by the SIM and data plan over a 10 year horizon.

2. Private LoRaWAN on the redefined LIPD band, upper sub block. Capex is gateway and backhaul. Opex is near zero. With 4 W EIRP gateways at three high sites and a typical regional path loss budget, 800 endpoints sit inside a single network with margin. The constraint is LBT. For hourly reads of small payloads this is not binding, but a firmware update push to the full fleet would have to be staged.

3. Wi SUN FAN on the same LIPD band. A mesh topology removes the gateway count constraint and adds resilience, at the cost of more complex network management and higher endpoint power consumption.

The class licensed options (2 and 3) are cheaper to operate and give the utility full control of the spectrum it depends on. They also carry an assumption: that the LBT and duty cycle constraints stay engineerable as more users move into the same band. The carrier option (1) transfers the spectrum risk to the carrier, at a recurring per endpoint cost.

The point of the example is that the right answer is no longer obvious. Five years ago the same utility would have stayed on 400 MHz UHF with a 2G fallback. The replan forces a real engineering decision, and the inputs to that decision are exactly the parameters that have changed: duty cycle rules, LBT thresholds, adjacent band cellular density, and the available filter front ends in commodity hardware.

---

What an RF engineer should actually do this quarter

If you operate or design networks that touch the 900 MHz region in Australia, the practical work breaks into four steps.

Audit

Pull every apparatus licence you hold or operate under in 803 to 960 MHz from the RRL. Tag each by licence type, expiry, and migration zone. Anything in 890 to 915 / 935 to 960 MHz that is not cellular needs a replacement plan.

Re coordinate

For colocated cellular and LPWAN sites, run an isolation calculation per port against the proposed emission masks. Where the calculation does not close, decide between added filtering, antenna repositioning, or moving the LPWAN service to a different sub block.

Re plan IoT links

For any class licensed deployment in 915 to 928 MHz, re run the link budget against the proposed sub block rules:

• The 920 to 926 MHz sub block at 4 W EIRP with LBT looks like the right home for most new LoRaWAN and Wi SUN deployments.
• The lower sub block remains the right home for very low duty cycle telemetry.
• The upper sub block looks best suited to RFID and very short range telemetry rather than general LPWAN use. That is an operational reading rather than a categorical exclusion.

Plan the 2G sunset

Inventory every cellular IoT module in the field by chipset. Anything that does not have a Cat M1 or NB IoT fallback needs a hardware refresh on a defined timeline, not when the device next fails.

---

Why this matters

The replan is not a regulatory inconvenience. It is the first time in two decades that the 900 MHz band has been reorganised end to end in Australia. For engineers working in IoT, mining, utilities, and rail, it changes the underlying assumptions of almost every network design decision in this region.

The networks that hold up under it will be the ones where the engineering team has already done the audit, run the numbers against the new masks, and chosen sub blocks and bearers deliberately. The networks that struggle will be the ones still operating against the band plan that existed before the consultation closed.

---

Sources and further reading

Use this article to frame the engineering questions, then verify the parameters against the primary documents. All are published by ACMA or on the Federal Register of Legislation:

• Radiocommunications (Low Interference Potential Devices) Class Licence, the in force instrument and amendments, on the Federal Register of Legislation and acma.gov.au.
• ACMA consultation papers and outcomes for the 900 MHz band re band and the LIPD review, on the ACMA consultations page.
• RALI (Radiocommunications Assignment and Licensing Instructions), in particular the MS series for fixed and land mobile coordination.
• Register of Radiocommunications Licences (RRL) for your own assignments and expiry dates.
• ACMA five year spectrum outlook for indicative timing of the migrations.

If a parameter in this article is load bearing for a design decision, treat the matching primary document as the authority, not this page.