A DU does not get a dedicated fibre to every radio. The fronthaul gateway is what puts many O-RUs onto one transport — aggregating and switching their eCPRI 7.2x streams, compressing them, and distributing the brutally precise timing every radio in the cluster depends on.
In a disaggregated RAN, the low-PHY split sends raw-ish samples between the O-RU and the DU over eCPRI — a lot of bandwidth, with hard timing. Giving every radio its own fibre to the DU does not scale. The fronthaul gateway aggregates a cluster of O-RUs onto a shared transport, switching the user, control, synchronisation and management planes to the right place.
It does the unglamorous work that makes fronthaul affordable: block floating-point compression to fit the link budget, plane-aware switching and QoS so control and sync are never starved by user data, and buffering across the transport.
The hard part is timing. The 7.2x split needs sub-microsecond alignment across every radio, so the gateway is a PTP / SyncE node that recovers and redistributes timing to the whole cluster. Get that wrong and the radios drift, TDD collides and beamforming falls apart.
| Parameter | Detail |
|---|---|
| Fronthaul | O-RAN 7.2x over eCPRI; aggregation and switching of U / C / S / M planes |
| Timing | PTP (IEEE 1588) and SyncE node; recovers and redistributes sub-microsecond timing to the cluster |
| Compression | Block floating-point (BFP) to fit the transport budget |
| Transport | Aggregates many O-RUs onto a shared fibre transport to one or more DUs; QoS per plane |
| Scale | A radio cluster onto one uplink; plane-aware switching and buffering |
| Management | Monitoring of timing, link health and per-plane traffic across the fronthaul |
| Use | Dense RU clusters, multi-RU sites, metro fronthaul aggregation |
The fronthaul gateway sits between the radios and the baseband wherever a cluster of O-RUs shares transport — dense urban sites, multi-band sites and metro fronthaul. It is the piece that turns many eCPRI links into one manageable transport, and the timing anchor the whole cluster trusts.
It pairs with the O-RAN radio units and AAUs it aggregates, the eCPRI IP core, and the DU it feeds.
We work on the gateway where it is hardest: the eCPRI plane-aware switching and QoS, the BFP compression, and above all the PTP / SyncE timing recovery and redistribution that the 7.2x split demands. Deliverables span architecture, FPGA/SoC implementation, timing characterisation and interop with O-RUs and DUs.
We can target the cluster size, the transport, and the timing budget against your deployment.
A node that aggregates and switches the eCPRI 7.2x fronthaul from several O-RUs onto a shared transport to the DU, compresses the traffic, and distributes the PTP/SyncE timing the radios depend on. It makes fronthaul scale beyond one fibre per radio.
Because a dedicated fibre per radio does not scale in dense deployments. The gateway puts a whole cluster of O-RUs onto one shared transport, switching the user, control, sync and management planes with QoS so none is starved.
The O-RAN lower-layer split that keeps low-PHY and RF in the O-RU and the rest in the DU. It carries a lot of bandwidth over eCPRI with hard timing, which is why aggregation, compression and precise synchronisation matter.
The 7.2x split needs sub-microsecond alignment across every radio. The gateway is a PTP (IEEE 1588) and SyncE node that recovers and redistributes timing to the cluster. If it drifts, TDD transmissions collide and beamforming breaks.
Block floating-point compression reduces the eCPRI payload so more radios fit on a transport link without losing the precision the PHY needs. It is a standard tool for meeting the fronthaul bandwidth budget.
Plane-aware eCPRI switching, BFP compression and sub-microsecond PTP/SyncE timing across a radio cluster are the work. Tell us the cluster and the transport.