Also known as: guard band, guardband
A guard band is a deliberately unused strip of spectrum placed between two adjacent channels, blocks, or services so that energy from one does not leak into the next.1 It is insurance against imperfect filters, finite roll-off, oscillator drift, and spurious emissions: by separating the occupied bandwidth of neighbours with a small vacant margin, a guard band trades a little spectral efficiency for a large reduction in adjacent-channel interference (ACI).
How it works
Every real transmitter’s spectrum has skirts — its power does not stop abruptly at the channel edge but rolls off over a finite transition, and small amounts of energy appear beyond the nominal band. A neighbouring receiver, meanwhile, cannot filter with infinitely steep sides, so it inevitably admits some energy from just outside its channel. The guard band exists so that where one transmitter’s residual skirt overlaps the next receiver’s filter skirt, both are already far down — turning what would be interference into negligible noise.
Concretely, channel spacing = occupied bandwidth + guard band. A regulator or standard picks a spacing that comfortably exceeds the emission’s occupied bandwidth; the surplus is the guard band. A signal with sharp pulse shaping and a small roll-off factor needs less guard band because its skirts are steeper; a sloppy or drifting emitter needs more. The guard band therefore couples directly to spectral efficiency: wider guards waste spectrum, tighter guards pack more channels but demand cleaner transmitters and better receiver selectivity.
In practice
Guard bands appear at every scale of the spectrum. Land-mobile channel plans leave a guard between each 12.5 kHz or 25 kHz channel; broadcast FM and TV allocations space stations so their masks do not collide; cellular and Wi-Fi blocks reserve guard bands at the edges of each operator’s allocation and between the block and its neighbours. In multicarrier systems like OFDM, unused edge subcarriers act as an internal guard band, and a guard interval (cyclic prefix) plays the analogous role in the time domain against inter-symbol interference. Whenever two services must coexist without coordination — such as a licensed band beside an unlicensed one — the guard band is what keeps them from stepping on each other.
Relevance to SDR
For a receive-only SDR, guard bands are why adjacent channels are separable at all: the vacant margin lets a channel filter isolate the wanted signal without dragging in the neighbour. When guard bands are thin and a nearby channel is much stronger, ACI can still bleed through, especially if the front end is not linear enough — a case where a narrower receive filter, more selectivity, or an attenuator to reduce the strong neighbour helps. Knowing the channel plan’s spacing and guard band tells you how tightly you can tune and filter before a neighbour intrudes.
GopherTrunk channelizes wideband captures into per-channel streams, and adequate guard bands in the source system’s plan are part of why those channels can be cleanly split. GopherTrunk does not create guard bands — they are a property of the transmitted spectrum — but its channelizer relies on them existing.
Sources
-
Guard band — Wikipedia, definition of the unused spectrum between channels and its role in preventing interference. ↩