Resource scheduling: How cellular networks allocate specific "resource blocks"

Key takeaways

How does a single cell tower serve hundreds or thousands of IoT devices at the same time, all sharing the same radio spectrum? The answer is resource scheduling — a system that divides available bandwidth into small, precisely defined units and assigns them to devices millisecond by millisecond. By the end of this lesson, you'll understand what resource blocks are, how the scheduler allocates them, and why this matters for your device's performance and power consumption.

What you'll learn

Why radio spectrum is a shared resource that must be managed

Cellular spectrum is finite, expensive, and shared. A typical LTE cell might have only 10 MHz or 20 MHz of bandwidth to serve every device in its coverage area. Unlike Wi-Fi, where devices compete for airtime in a "listen and shout" model, LTE uses a centrally controlled approach: the eNodeB acts as a traffic controller, deciding exactly which device uses which slice of spectrum at which moment. This is a major reason cellular networks are more reliable and predictable than Wi-Fi — there are no collisions, no hidden-node problems, and no airtime wasted on failed transmissions.

What a resource block is on the time-frequency grid

LTE divides bandwidth across two dimensions — frequency and time — forming a two-dimensional grid whose smallest schedulable unit is the resource block.

• In frequency: each resource block is 180 kHz wide, made of 12 subcarriers spaced 15 kHz apart. A 10 MHz cell has 50 resource blocks; a 20 MHz cell has 100.
• In time: one radio frame is 10 milliseconds long and contains 10 subframes of 1 ms each. Each subframe is a Transmission Time Interval (TTI), the basic scheduling period. A resource block spans one 0.5 ms slot, containing 7 OFDM symbols.

So one resource block is a rectangle 180 kHz wide × 0.5 ms long, holding 12 × 7 = 84 resource elements — and each resource element (one subcarrier for one OFDM symbol) is the smallest unit that can carry data.

How the eNodeB scheduler decides who gets resources

The scheduler runs every TTI — every single millisecond — and weighs four factors in each decision:

• Pending data. It maintains a queue of transmissions waiting for each connected device.
• Channel quality. Each device reports its conditions using Channel Quality Indicator (CQI) reports. A strong signal allows a more efficient modulation scheme like 64-QAM (more data per resource block); a weak signal needs a more robust but less efficient scheme like QPSK.
• Quality of Service. Latency-sensitive traffic (alarms) is prioritized over delay-tolerant traffic (routine telemetry).
• Fairness. The scheduler balances efficiency against making sure every device gets some service.

Common scheduling algorithms include Round Robin (equal time regardless of signal — simple but inefficient), Maximum Throughput (always serves the best channel — efficient but can starve weak devices), and Proportional Fair (the most widely used, balancing throughput and fairness by weighing each device's current channel quality against its average).

How uplink and downlink scheduling differ

Scheduling works differently in each direction:

• Downlink (network to device) uses OFDMA (Orthogonal Frequency Division Multiple Access). The eNodeB can assign different resource blocks to different devices within the same subframe — Device A at the top of the frequency range and Device B at the bottom, in the same millisecond.
• Uplink (device to network) uses SC-FDMA (Single Carrier Frequency Division Multiple Access). Each device must be assigned a contiguous block of subcarriers. This constraint exists because SC-FDMA has a lower peak-to-average power ratio, which lets the power amplifier in a battery-powered modem run more efficiently.

For IoT, the uplink usually matters most because most traffic is sensor data flowing from device to cloud — and the contiguous-block requirement makes uplink scheduling a harder optimization problem than downlink.

Why resource scheduling matters for your IoT device

Even though scheduling runs automatically in the eNodeB, it directly shapes device behavior:

• Throughput limits. The number of resource blocks assigned determines your throughput. In a congested cell, your device may get fewer blocks, meaning lower throughput and longer transmissions. NB-IoT takes this to the extreme, operating in a single 180 kHz resource block, which is why its peak rate is limited to 26–127 kbps versus LTE-M's roughly 1 Mbps over a wider 1.4 MHz allocation.
• Power consumption. The longer the radio stays active to finish a transmission, the more battery it uses. Fewer resource blocks (from congestion or poor signal) mean longer transmissions and more drain — under poor signal with multiple repetitions, energy per bit can rise sharply, potentially halving expected battery life.
• Coverage extension through repetitions. For cell-edge devices with weak signal, LTE-M and NB-IoT repeat the same data across multiple resource blocks so the receiver can combine them to decode it. This improves the link budget (up to 164 dB maximum coupling loss for NB-IoT) but consumes more blocks and power.
• Scheduling Request latency. An idle device must first request resources before sending, which adds latency — typically 10–15 ms in normal conditions. This matters for time-sensitive applications but is negligible for routine telemetry.

Lesson 2.3 Self-assessment

Frequently asked questions