What is the difference between licensed and license-free spectrum?

Licensed spectrum is assigned to a named operator who pays for exclusive or protected use, cellular bands are the classic example. License-free (license-exempt) spectrum is open to any device that meets a published set of technical rules, with no individual permit and no protection from interference. SRD and ISM bands belong to this second category. The trade-off is freedom of access against the obligation to tolerate interference and to share the band with everyone else.

What is the difference between EIRP and ERP?

Both express radiated power, but against different references. EIRP (Equivalent Isotropically Radiated Power) is referenced to an ideal isotropic antenna. ERP (Effective Radiated Power) is referenced to a half-wave dipole. The dipole already has about 2.15 dBi of gain over isotropic, so the same physical emission reads about 2.15 dB higher in EIRP than in ERP. A limit always states which reference it uses, and you must convert before comparing two figures.

Which ETSI standard applies to my band?

As a first orientation, EN 300 220 covers SRD operation below 1 GHz (the 433 MHz and 868 MHz bands), EN 300 328 covers wideband data transmission in the 2.4 GHz band (Wi-Fi, Bluetooth, Zigbee), and EN 301 893 covers the 5 GHz RLAN band. These are the harmonised standards that give presumption of conformity to RED article 3.2 in the EU. The exact standard depends on band, bandwidth and technology, so confirm against the latest list of harmonised standards.

Why does 868 MHz work in Europe but not in the United States?

Because the sub-GHz SRD allocations differ by region. The EU SRD band sits around 868 MHz under CEPT/ECC ERC Recommendation 70-03, while the US equivalent sits around 902 to 928 MHz (commonly called the 915 MHz band) under FCC Part 15. A device tuned to 868 MHz has no legal allocation in the US, and a 915 MHz device has none in the EU. A product sold on both markets needs a band-switchable radio and two separate compliance dossiers.

What is duty cycle and why does it limit my throughput?

Duty cycle is the fraction of time a transmitter is allowed to be active over a defined window, often one hour. In several EU sub-GHz sub-bands the regulation caps duty cycle (for example a small percentage per hour) to keep the shared band usable for everyone. A low duty-cycle cap directly limits how much data you can send, so for chatty applications you either pick a sub-band with a higher allowance, use listen-before-talk where permitted, or move to a different band.

What are DFS and TPC in the 5 GHz band?

DFS (Dynamic Frequency Selection) and TPC (Transmit Power Control) are mandatory mitigation techniques on parts of the 5 GHz band that are shared with radar. DFS makes the device listen for radar and vacate the channel if it detects one. TPC reduces transmit power to the minimum needed, lowering interference to satellite and radar services. EN 301 893 in the EU and FCC Part 15.407 in the US both require these mechanisms on the relevant 5 GHz sub-bands.

Does a certified radio module let me skip band compliance?

No. A pre-certified module carries its own grant, but the moment you integrate it into a product with your own antenna, enclosure and firmware you become responsible for the radiated result. Antenna gain, placement and host noise all affect EIRP, spurious emissions and channel access. The integrator must still document that the finished product respects the band rules, which is exactly the test scope set by EN 300 220, EN 300 328 or EN 301 893.

Radio spectrum: SRD and license-free bands

Author Hugues Orgitello Last updated 28 May 2026 ~11 min read

Guide · Radio spectrum orientation

Before a wireless product can exist, it must answer one question: can I legally transmit on this frequency, at this power, in this region? Most consumer and industrial radios never touch licensed spectrum. They live in the license-free Short Range Device and ISM bands, where access is open to anyone who respects a published rulebook. This page maps those bands, the parameters that govern them (power, duty cycle, channel access), and the standards that turn the rules into a test plan. It is the orientation layer under the RED pillar: once you know which band and which standard apply, the rest of the certification path follows.

Licensed versus license-free spectrum

The radio spectrum is a finite shared resource, and every country administers it. Two regimes coexist.

Licensed spectrum is assigned to a named holder, usually after auction or allocation, who pays for the right and enjoys protection from interference. Cellular operators, broadcasters and many satellite services work here. A device cannot transmit in licensed spectrum without the holder's authorisation, which is why a phone relies on the operator's network rights rather than its own.

License-free spectrum (also called license-exempt) is open to any equipment that complies with a published set of technical conditions. There is no individual permit and no fee, but there is also no protection: a license-exempt device must accept interference from others and must not cause harmful interference itself. This is the bargain that makes Wi-Fi, Bluetooth, Zigbee, LoRa and countless remote controls possible. The rulebook is what keeps the band usable for everyone.

The practical consequence: in licensed bands you negotiate access; in license-free bands you earn access by conformity. The rest of this guide is about the second case.

SRD and ISM bands at a glance

Two overlapping terms appear constantly. ISM (ISM bands, Industrial, Scientific and Medical) historically designated bands set aside for non-communication uses such as microwave heating, where communication devices operate on a no-protection basis. SRD (Short Range Device) is the regulatory category for low-power radios designed for short range, covered in the EU by CEPT/ECC ERC Recommendation 70-03. In everyday use the two terms blur, the 2.4 GHz band is both an ISM band and the home of countless SRDs, but the distinction matters when you read a regulation, because the allowance attaches to a specific sub-band and use.

The table below orients the main license-free bands used in electronics. Treat the figures as the band's character, not as legal limits to quote verbatim, the binding numbers live in the standards and recommendations cited at the end.

Band	Region	Typical use	Primary EU standard	Primary US rule
433 MHz	EU (SRD)	Remote controls, simple telemetry	EN 300 220	Part 15.231 / 15.240
868 MHz	EU (SRD)	LoRa, sub-GHz IoT, metering	EN 300 220	not allocated (use 915 MHz)
902 to 928 MHz (915 MHz)	US (ISM)	LoRa, sub-GHz IoT, metering	not allocated (use 868 MHz)	FCC Part 15.247
2.4 GHz	Global	Wi-Fi, Bluetooth, Zigbee, Thread	EN 300 328	Part 15.247
5 GHz	Global	Wi-Fi (RLAN)	EN 301 893	Part 15.407
sub-GHz (general)	Region-specific	Long-range, low-rate IoT	EN 300 220	Part 15 Subpart C

The single most important takeaway from this table: 868 MHz is European, 915 MHz is American. They are not interchangeable. A sub-GHz product intended for both markets carries two radio configurations, two test campaigns and two declarations. The 2.4 GHz band is the only one with a genuinely global character, which is why most multi-region products lean on it.

The key parameters

Every license-free band is governed by the same family of parameters. Understanding them is what lets you read any band's rule and predict how it constrains a design.

EIRP versus ERP

Power limits are the headline constraint, but a number means nothing without its reference. Two references dominate.

EIRP (Equivalent Isotropically Radiated Power) is referenced to an ideal isotropic radiator that emits equally in all directions.
ERP (Effective Radiated Power) is referenced to a half-wave dipole.

A dipole has roughly 2.15 dBi of gain over isotropic, so the same physical emission reads about 2.15 dB higher when expressed as EIRP than as ERP. Limits are stated one way or the other, never both, so the first thing to check on any band is which reference the limit uses. Comparing an EIRP figure to an ERP figure without converting is one of the most common spectrum-planning errors.

The radiated figure also depends on antenna gain. A transmitter delivering a modest conducted power into a high-gain antenna can exceed an EIRP limit, while the same conducted power into an inefficient antenna may sit well under it. This is why power limits are usually expressed as radiated power: the regulator cares about what leaves the product, not what enters the antenna connector.

Transmit power limits

Each band, and often each sub-band within it, sets a maximum radiated power. Common reference points include 25 mW (about 14 dBm) for many EU SRD sub-bands and 100 mW EIRP (20 dBm) for the 2.4 GHz band in the EU. The 5 GHz band uses different ceilings per sub-band, with the higher allowances tied to the DFS and TPC obligations described below. Always read the exact figure from the applicable standard rather than a remembered round number, because allowances differ between sub-bands and between regions.

Duty cycle restrictions

In several EU sub-GHz sub-bands, raw power is not the only brake. Duty cycle caps the fraction of time a transmitter may be active over a defined window, often one hour. A duty-cycle limit shares the band in time the way a power limit shares it in space. It directly bounds throughput: a low duty-cycle allowance means a device can only send for a small slice of each hour, which is why long-range sub-GHz protocols are built around short, infrequent messages. Where the regulation permits, listen-before-talk can replace or relax a fixed duty cycle.

Channel access and listen-before-talk

LBT (Listen Before Talk) is a channel-access discipline: the device listens for energy on the channel and only transmits if the channel is clear. It is a politeness mechanism that lets many devices coexist without a central coordinator. Some bands and modes require LBT (often alongside Adaptive Frequency Agility), others accept a fixed duty cycle instead, and a few allow either. The choice affects both the radio design and the test scope, because the standard verifies that the implemented mechanism behaves as claimed.

Occupied bandwidth and spectral mask

Two further parameters complete the picture. Occupied bandwidth is the span of frequency the signal actually uses, and most bands cap it so that a single device cannot monopolise the band. A wider channel buys throughput but consumes more of the shared resource and can push energy toward the band edges. The spectral mask (or out-of-band emission limit) defines how quickly the emission must fall off outside the assigned channel, protecting neighbouring channels and adjacent services. A transmitter that is well within its in-band power limit can still fail on the mask if its modulation has lazy skirts or its power amplifier is driven into non-linearity. These two measurements appear in every radio test plan and are a common cause of a first-pass failure, so they deserve attention at the design stage, not at the lab.

The 5 GHz special case: DFS and TPC

The 5 GHz RLAN band deserves its own section because parts of it are shared with radar (weather radar, military and aeronautical systems). To protect those incumbents, two mitigation techniques are mandatory on the affected sub-bands.

DFS (Dynamic Frequency Selection) requires the device to monitor for radar signatures and, on detecting one, to stop transmitting on that channel and move elsewhere within a defined time. The channel is then avoided for a set period before it can be reused.
TPC (Transmit Power Control) requires the device to use no more power than needed, reducing the interference footprint toward satellite and radar services.

Both EN 301 893 in the EU and FCC Part 15.407 in the US mandate these mechanisms on the relevant 5 GHz sub-bands. They are not optional features, they are conformity requirements, and DFS testing (radar-waveform detection, channel-move time, non-occupancy period) is a substantial and failure-prone part of a 5 GHz test campaign. A product that wants the higher-power 5 GHz sub-bands must implement and prove both.

Regional differences: EU versus US

The same physical band can carry different rules on either side of the Atlantic. The table below contrasts the two regulatory families for the bands that matter most.

Aspect	European Union	United States
Framework	RED 2014/53/EU, article 3.2	FCC Part 15
SRD reference	ERC Recommendation 70-03	Part 15 Subpart C
Sub-GHz band	868 MHz	902 to 928 MHz (915 MHz)
Sub-GHz standard	EN 300 220	FCC Part 15.247 / 15.231
2.4 GHz standard	EN 300 328	Part 15.247
5 GHz standard	EN 301 893	Part 15.407
Authorisation route	self-declaration or Notified Body	SDoC or TCB certification

The European reference for license-free SRDs is ERC Recommendation 70-03, maintained by CEPT/ECC. It consolidates the SRD frequency allocations and conditions (power, duty cycle, channel access) that the harmonised ETSI standards then translate into testable requirements. National administrations implement it, so there can be minor country variations, but 70-03 is the single document to start from for any EU SRD question.

In the US, the equivalent body of rules is 47 CFR Part 15, with Subpart C governing intentional radiators (devices designed to emit RF). The frequently cited sections are 15.247 (the 2.4 GHz and 915 MHz digital-modulation and frequency-hopping band) and 15.407 (the 5 GHz U-NII band, including its DFS and TPC obligations). The US authorisation route, SDoC or TCB certification depending on the device, parallels the EU split between self-declaration and Notified Body, covered in our guide on the certification routes.

From band to RED article 3.2 standard

In the EU, the RED sets three essential requirements. Article 3.2, the efficient use of spectrum, is the one that radio spectrum planning feeds directly. The chain is straightforward.

Identify the band and technology. Sub-GHz SRD, 2.4 GHz wideband, 5 GHz RLAN, each maps to a different harmonised standard.
Select the article 3.2 standard. EN 300 220 for sub-GHz SRD, EN 300 328 for 2.4 GHz wideband data, EN 301 893 for 5 GHz RLAN. Applying the standard in full gives presumption of conformity to article 3.2.
Confirm full applicability. If the standard does not cover your modulation or operating mode, the presumption is incomplete and a Notified Body may be needed, the boundary explained in self-declaration vs Notified Body.
Plan the test campaign. The chosen standard defines the radio measurements: power, occupied bandwidth, spurious emissions, channel access, and for 5 GHz the DFS and TPC tests.

Article 3.2 is not the whole RED. Article 3.1 covers safety and health (including field exposure, see the SAR procedures guide), and article 3.3 covers cybersecurity. But the spectrum question, which band and which 3.2 standard, is the one that flows from the choices made on this page.

Consequences for product design

Spectrum rules are not a box ticked at the end. They shape the radio architecture from the first schematic.

Band choice drives antenna and range. Sub-GHz bands propagate further and penetrate buildings better than 2.4 GHz, at the cost of lower data rates and tighter duty-cycle limits. The application's range, throughput and battery budget decide the band, and the band decides the rulebook.
Power and antenna gain are a single budget. Because limits are radiated (EIRP or ERP), antenna gain counts against your conducted power. A high-gain antenna can push a compliant transmitter over the limit. Plan the link budget and the regulatory budget together.
Duty cycle constrains the protocol. If the band caps duty cycle, the application protocol must fit inside that envelope, short frames, low repetition, or a band with more headroom. Retrofitting a chatty protocol onto a low duty-cycle band is a frequent and painful mistake.
Channel access must be implemented and proven. If the band requires LBT, the firmware must implement it correctly and the test lab must confirm it. The same applies to 5 GHz DFS and TPC, where the radar-detection behaviour is verified against defined waveforms.
Multi-region products carry multiple radios. Because 868 MHz and 915 MHz do not overlap, a worldwide sub-GHz product needs a band-switchable design and a separate dossier per region. Designing for one market and bolting on the other later is far more expensive than planning for both up front.

These constraints scale up at the protocol level. Certification programmes such as LoRaWAN, Wi-Fi and Bluetooth sit on top of the regulatory band rules: the alliance qualification proves interoperability, while the band standards prove lawful spectrum use. Both are needed, and they are distinct.

A worked orientation

Case A: a sub-GHz IoT sensor for the EU and US

The application needs long range and low power, so sub-GHz is the natural band. In the EU it transmits at 868 MHz under EN 300 220, respecting the relevant sub-band power and duty-cycle conditions from ERC Recommendation 70-03. In the US the same product must move to the 902 to 928 MHz band under Part 15.247. Outcome: one product, two radio configurations, two test campaigns, two declarations. The band rules, not the silicon, drive the project plan.

Case B: a 2.4 GHz Bluetooth wearable

The 2.4 GHz band is global, so a single radio configuration serves both markets. In the EU the radio test follows EN 300 328; in the US it follows Part 15.247. Frequency-hopping and adaptivity satisfy the channel-access expectations of both. Outcome: one radio, two reports, one product worldwide. This is why 2.4 GHz dominates consumer wireless.

Case C: a 5 GHz Wi-Fi access point

To reach the higher-power 5 GHz sub-bands, the design must implement DFS and TPC. The test campaign under EN 301 893 (EU) and Part 15.407 (US) includes radar detection, channel-move time and non-occupancy, on top of the usual power and emission measurements. Outcome: a heavier and more failure-prone radio campaign, planned from the start because DFS cannot be added late.

For the receiver-side measurements that complete a radio dossier, see the guide on spurious emissions, receiver blocking and immunity tests. For the full EU radio path, start from the RED pillar.