LTE to 5G NR: what changes for the optimiser

What actually changes

Much of 5G NR is familiar. The KPI framework transfers directly, the elimination order for troubleshooting is unchanged, and a link budget is still a link budget. What changes is the physical layer underneath, the way the network is measured, and the fact that for several years most deployments run two radio technologies at once.

Flexible numerology

LTE has one subcarrier spacing. NR has several, selected by a numerology index, and each doubling of the spacing halves the slot duration. This is the change with the widest downstream consequences.

• Wider spacing shortens the slot, which reduces latency and suits high-band deployment.
• Wider spacing tolerates more Doppler and phase noise, which is what makes millimetre-wave operation practical at all.
• Narrower spacing gives a longer cyclic prefix, which is what makes large cells and heavy multipath workable.

The optimiser's consequence: numerology is a coverage and latency decision, not a throughput one. Comparing cells on different numerologies without accounting for it produces conclusions that do not hold.

Measurement moves to the synchronisation block

LTE transmits a cell-specific reference signal continuously across the whole carrier, which is convenient to measure and expensive in energy and interference. NR removes it. Measurement is instead based on the synchronisation signal block, transmitted periodically and in a beam.

Three practical consequences follow, and all three catch out engineers moving from LTE:

• Measurements are per beam, not per cell. A cell has several synchronisation beams, and a device measures the ones it can hear. Cell-level coverage is an aggregate, and hides beam-level holes.
• Measurement is periodic. The timing configuration determines when a device may measure at all, so an incorrect configuration produces mobility failures that look like coverage problems.
• An idle carrier is quiet. Without an always-on reference signal, an unloaded NR cell radiates far less than an unloaded LTE cell. Interference measured at low load will understate interference at busy hour.

Beamforming and massive MIMO

Large antenna arrays — commonly 32 to 256 elements — allow energy to be steered toward individual devices rather than sprayed across a sector. The gain is twofold: more received power for the target, and less interference for everyone else.

The planning consequence is that a beamformed cell no longer has a single fixed footprint. Coverage depends on the beam set in use, so a static prediction is less meaningful than it was in LTE, and beam-level measurement matters more than sector-level averages.

Spatial multiplexing also becomes the primary capacity lever. Where LTE capacity came mainly from bandwidth and modulation, NR adds independent spatial streams to the same time and frequency resource — which is why massive MIMO gain shows up in cell capacity rather than in single-user peak rate.

Non-standalone and dual connectivity

Most 5G networks launched as non-standalone: the device anchors on LTE for control and adds an NR carrier for user plane throughput. LTE carries mobility and signalling; NR carries data.

For the optimiser this means the LTE layer determines 5G accessibility and retainability. A device that cannot hold the anchor cannot use the NR carrier at all, so a 5G availability problem is very often an LTE coverage problem. Diagnosing it on the NR carrier alone will not find it.

The specific KPIs that matter in this mode are secondary node addition success, the proportion of time a device retains the NR leg, and the rate of secondary node change failure. Threshold tuning for adding and releasing the NR carrier trades throughput against stability in exactly the way handover offsets do in LTE.

Standalone and uplink coverage

Standalone operation removes the LTE anchor: NR carries both control and user plane against a 5G core, which is what enables slicing, lower latency and voice over NR.

The constraint that dominates standalone planning is uplink coverage. A mid-band NR carrier has ample downlink reach, but a handheld device cannot match the base station's transmit power, so the uplink fails first. Downlink coverage consistently overstates usable range.

Two mechanisms address this directly:

• Supplementary uplink — pairing the mid-band downlink with a low-band uplink carrier, so the uplink gets the propagation advantage of the lower frequency.
• Uplink and downlink decoupling — allowing the uplink to be served by a different carrier or cell from the downlink, matching each direction to where it performs best.

Both are worth checking before concluding that a standalone cell needs more sites.

RAN slicing

Network slicing partitions one physical network into logical networks with independent characteristics — a low-latency slice, a high-throughput slice, a massive-device slice — each identified by a slice identifier the device signals during connection.

In the radio network a slice is enforced through resource partitioning and scheduling priority. Two design questions govern it, and they are commercial as much as technical:

• Hard or soft partitioning. Reserving resources guarantees a slice its capacity but wastes it when the slice is idle. Priority-based sharing uses resources efficiently but only guarantees relative treatment, not absolute capacity.
• What is actually being promised. A slice with a throughput commitment has to be dimensioned for it, which means the guarantee has a site count attached. A slice sold without that arithmetic is a commitment the radio network cannot keep.

Slicing also changes reporting: KPIs must be produced per slice, since an aggregate cell KPI will average a failing slice against a healthy one and show neither.

Voice: two strategies

Voice on 5G resolves to one of two approaches, and which one applies depends on whether the network is standalone.

• Fallback to LTE — the standard approach on non-standalone and on early standalone networks. A voice call moves the device to the LTE layer and is carried as VoLTE. Simple and reliable; the cost is the setup delay of the transition, which shows up directly in post-dial delay.
• Voice over NR — voice carried natively on the NR carrier, requiring standalone operation and a 5G core. Removes the fallback delay and keeps the device on 5G, but requires NR coverage good enough to carry a delay-sensitive service everywhere voice is expected.

The VoLTE material in this Knowledge Centre applies to both: the bearer model, the quality measurements and the troubleshooting sequence carry over. Where voice falls back to LTE, the LTE voice layer remains the thing that has to work.

Carrier aggregation and peak rate

Aggregating carriers remains the most direct way to raise peak rate, and NR extends it across bands and across the two technologies at once. The arithmetic is simple — capacity scales with total bandwidth — but two caveats apply in practice.

First, aggregation raises peak rate for devices that are already in good conditions; it does nothing for a device at the cell edge, which is where complaints originate. Second, each added carrier has its own coverage footprint, so the aggregated rate is only available where every carrier is available — a much smaller area than the primary carrier's own footprint.