Giving every radio unit its own fibre back to the distributed unit is the simple answer and frequently the unaffordable one. Cascading them — each unit passing traffic through to the next — removes most of that fibre. What it does not remove is the bandwidth, the timing budget and the failure domain, and those three between them decide how long a chain can be.
| Topology | How it works | Trade |
|---|---|---|
| Star | A dedicated fronthaul link from the distributed unit to each radio unit | Simplest timing and the smallest failure domain, at the cost of one fibre run per radio |
| Chain (daisy chain) | Each radio unit passes traffic through to the next along a single path | Dramatically less fibre; bandwidth is shared and everything downstream of a break goes with it |
| Ring | A chain closed back to the distributed unit, so traffic can travel either way | Survives a single break by reversing direction, at the cost of a return path and protection logic |
Cascading is a civil engineering decision expressed in network topology. It is chosen where fibre is expensive, disruptive or physically impossible to run, not because it is technically preferable.
The geometry that suits cascading is a line rather than an area. Road and rail tunnels, station platforms, long industrial corridors, mine drifts, and building risers all have the same shape: coverage needed along a path, and civil works that make a dedicated run to each radio disproportionately expensive.
In those settings a chain of modest radios is frequently cheaper to install and easier to permit than a star of the same count, and the shared bandwidth is tolerable because per-radio traffic is low.
Where it fits badly is dense capacity. A stadium, a large venue or an urban macro cluster wants high per-radio throughput and a small failure domain, and both argue against a chain. The honest answer in those cases is that fibre is expensive and worth it.
Connecting radio units in a chain, where each unit passes fronthaul traffic through to the next, rather than giving every unit a dedicated link back to the distributed unit. It removes most of the fibre at the cost of sharing bandwidth and enlarging the failure domain.
It depends on which of three limits binds first: aggregate fronthaul bandwidth on the first hop, accumulated timing error at the far end of the chain, and the size of the failure domain that is acceptable. Bandwidth usually decides it, and the answer falls sharply as per-radio bandwidth and antenna count rise.
Because synchronisation propagates hop by hop and each hop adds delay and a little uncertainty. The last radio has the least margin, so the timing budget is calculated for that position rather than for the average.
Everything downstream of the break loses service. Where that is unacceptable the chain is closed into a ring so traffic can reverse direction, which costs a return path and protection logic but survives a single break.
Where capacity per radio is high and the failure domain must stay small — stadiums, large venues, dense urban clusters. In those cases the shared bandwidth and the chain-wide outage risk outweigh the fibre saving.
Written for engineers. Share it with one.
The hardware in the chain.
TECHNOLOGYRU, DU, CU and the haul between.
KNOWLEDGESSB, TRPs and beam mobility.
PRODUCTTiming distribution along the chain.
SOLUTIONAnother long, thin deployment.
INDUSTRYDeployment economics.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.