Network capacity is the maximum amount of traffic that a telecommunications network, or one part of it, can carry in a given time while still meeting its quality targets. In network performance, capacity describes the shared resource behind every individual measurement: it is the total that all users of a cell, link or node divide among themselves, usually expressed in megabits or gigabits per second for data and in erlangs for voice.
Capacity is different from the speed a single customer sees. A speed test reports the throughput of one connection at one moment, whereas capacity sets the ceiling for everyone together. When demand approaches that ceiling, congestion appears, latency rises and per-user throughput falls, even if signal strength is excellent.
What determines the capacity of a mobile cell
The capacity of a radio cell follows from three multiplying factors: the amount of spectrum, the efficiency with which that spectrum is used and the number of independent spatial streams. Shannon’s theorem sets the theoretical limit: capacity grows linearly with bandwidth but only logarithmically with signal quality, which is why operators value additional spectrum so highly.
Spectral efficiency, measured in bits per second per hertz, depends on modulation, coding and the SINR of each user. Users at the cell edge need robust, low-order modulation and therefore consume far more radio resources per megabit than users near the antenna. The table shows the main levers and their typical effect.
| Factor | Typical values | Effect on capacity |
|---|---|---|
| Channel bandwidth | 10-20 MHz in LTE, up to 100 MHz in 5G mid-band | Roughly proportional |
| Modulation order | QPSK up to 256-QAM or 1024-QAM | More bits per symbol at high SINR |
| MIMO layers | 2×2 to 64T64R massive MIMO | Several streams on the same spectrum |
| Average SINR | Below 0 dB at the edge, above 20 dB near the site | Sets usable modulation |
| Sectorization | 3 sectors per site, sometimes 6 | Reuses spectrum in each direction |
Because these factors interact, a nominal peak data rate of several gigabits per second describes ideal conditions for one user. Real average cell throughput is typically a fraction of that figure, often between one quarter and one half, depending on user distribution, interference and traffic mix.
Capacity, throughput and related measures
Capacity is a design and planning quantity, while most everyday metrics describe what one user experiences at a particular moment. Confusing them leads to wrong conclusions, such as blaming coverage for a problem that is really a shortage of capacity. The comparison below separates the closely related terms.
| Metric | What it measures | Scope |
|---|---|---|
| Network capacity | Total traffic the resource can carry | Cell, link or whole network |
| Peak data rate | Theoretical maximum for one user | Single device, ideal conditions |
| Download speed | Achieved throughput toward the user | Single connection, one moment |
| Signal strength | Received power, for example RSRP | Radio link quality, not load |
| Network congestion | Demand exceeding available capacity | Condition of a shared resource |
A useful rule of thumb follows from this distinction: strong signal with slow speed usually points to a capacity or backhaul limit, while weak signal with slow speed points to coverage. RSRQ and SINR help separate the two, because they fall when neighboring cells are busy even if RSRP stays high.
Capacity beyond the radio interface
Radio is only the first link in a chain, and the narrowest link sets the end-to-end capacity. Each site connects through backhaul (microwave, fiber or copper) to aggregation routers, then to the core network and finally to internet peering and transit. Any of these points can become the bottleneck.
Fixed networks follow the same logic. In passive optical networks, a single fiber feeder is shared among many homes through a splitter, and the split ratio determines how much capacity each subscriber can count on at peak times. Cable networks share capacity per node in a similar way.
- Backhaul: a 1 Gbit/s microwave link cannot serve a 5G site that can deliver 3 Gbit/s on the radio side.
- Aggregation and core: routers and user plane functions need headroom for growth and failover.
- Peering and transit: interconnection with content networks limits traffic during major live events.
- Shared fixed access: GPON and cable segments divide one pipe among dozens of households.
How operators plan and expand capacity
Operators plan capacity around the busy hour, the period of the day with the highest load, rather than around averages. A common practice is to trigger an upgrade when a cell or link reaches a utilization threshold, often in the range of 60-80 percent of busy-hour capacity, because latency and packet loss grow sharply as load approaches 100 percent.
The choice of upgrade depends on cost, lead time and where the bottleneck sits. The decision table summarizes the usual options from the cheapest to the most expensive.
| Situation | Typical measure | Reason |
|---|---|---|
| Spare licensed spectrum available | Add a carrier or widen the channel | Fast gain without new sites |
| High load, good SINR | Upgrade to massive MIMO | Serves more users on the same spectrum |
| Dense hotspot demand | Add small cells | Reuses spectrum over short distances |
| Radio idle, users still slow | Upgrade backhaul to fiber | Removes transport bottleneck |
| Legacy traffic on old layers | Refarm 3G spectrum to 4G or 5G | Higher spectral efficiency |
Operators also manage demand. Quality of service rules prioritize voice and critical traffic, and some tariffs apply mobile data throttling after a usage cap. These tools do not create capacity; they decide who gets it when it is scarce.
Common misconceptions about network capacity
Marketing figures and individual speed tests shape public perception, yet both describe a single moment for a single user. Several misunderstandings follow from this view.
- A faster generation does not remove congestion: a 5G cell with many users can still be slower per user than a lightly loaded 4G cell.
- Full signal bars do not mean free capacity: bars reflect received power, not how many people share the cell.
- Capacity is not evenly divided: schedulers allocate resources by radio conditions and fairness rules, so edge users receive less.
- One speed test is not a capacity measurement: repeated tests across times of day reveal the busy-hour pattern.
A plain-language summary
Network capacity is like the number of lanes on a highway. The speed limit tells how fast one car may drive, but the number of lanes decides how many cars can move at once before a traffic jam forms.
A mobile cell is shared by everyone nearby. At three in the morning, one phone may get very high speed. At a stadium during a match, thousands of phones divide the same lanes, and each one gets much less. More capacity means more lanes, not a higher speed limit.
An example from practice
A regional operator notices that a cell serving a busy commuter train station reaches 85 percent utilization every weekday between 7:30 and 9:00. Customers report slow downloads even though their phones show strong signal. Measurements confirm good RSRP but falling SINR and long scheduling queues, which point to a capacity problem rather than a coverage gap.
The engineering team first adds a second 20 MHz carrier from spare mid-band spectrum, which roughly doubles the radio capacity of the sector. Monitoring then shows that the 1 Gbit/s microwave backhaul has become the new limit, so the site is connected to fiber three months later. After both changes, busy-hour utilization falls to about 45 percent, and median download speed during the morning peak triples.