A private 5G network is shaped by the spectrum it is allowed to use before a single radio is chosen. Whether the band is licensed, shared or unlicensed, and whether it sits low, mid or millimetre-wave, decides the coverage, the capacity and the cost — so the regulatory regime is the first design input, not a detail to settle at the end.
| Band / regime | Strength | Cost / constraint |
|---|---|---|
| Low band (sub-GHz) | Long range, penetrates walls | Little bandwidth, so lower capacity |
| Mid band (3–4 GHz) | Capacity and coverage balanced | The contested, popular middle ground |
| mmWave (24 GHz+) | Very high capacity | Short range, needs line of sight, dense cells |
| Licensed | Interference-protected, guaranteed | Cost and, often, an operator relationship |
| Shared / local-licensed | Low barrier, dedicated locally | Availability and rules vary by country |
| Unlicensed | No fee, fast to start | Shares the band; no interference protection |
The single fact that governs radio planning is that low frequencies travel far and through walls, and high frequencies carry more data but stop at the first obstacle. A sub-gigahertz band covers a large site or a whole facility from few cells but offers modest bandwidth; millimetre-wave delivers enormous capacity but needs a dense grid of cells and a clear line of sight. Mid band is popular for private networks precisely because it compromises — enough reach to be practical, enough capacity to be useful — which is why the contest for it is fierce.
Layered on top of the physics is the licensing regime, and it is the part most likely to change the whole design. A licensed band is protected from interference and predictable, at the cost of a lease or an arrangement with an operator. A shared or locally-licensed regime — where a regulator lets a site use a band on its own premises — has lowered the barrier to private networks dramatically, but its availability and rules differ country to country. Unlicensed spectrum is free and immediate, but the network shares it with everyone else and has no protection when a neighbour switches something on.
Because of that, spectrum is the first design input, not the last. The band decides how many cells the site needs, where they go, what they cost to install and power, and how the network behaves under interference. Designing the application and the radios first and choosing spectrum afterwards is how projects discover, late, that the band they can actually get will not cover the site the way they assumed.
Because frequency decides the physics of the network. Low bands travel far and through walls but carry less data; high bands carry far more data but stop at obstacles and need line of sight. That single trade-off sets how many cells a site needs, where they go and what they cost, so the band is not a detail — it shapes the coverage, capacity and capital cost of the whole network.
It is a regime in which a regulator lets a site use a band on its own premises without a full national operator licence, sometimes after registering or coordinating to avoid clashes. It has made private 5G far more accessible by lowering the barrier to dedicated spectrum, but exactly which bands are offered and under what rules varies significantly by country, so it has to be checked locally rather than assumed.
Yes, and it is the fastest and cheapest way to start because there is no fee or licence. The trade-off is that the network shares the band with everyone else using it and has no interference protection, so performance can degrade unpredictably when a neighbour becomes active. It suits some deployments, but where reliability matters a licensed or shared-licensed band, which offers protected or dedicated access, is usually worth the cost.
Written for engineers. Share it with one.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.