Carrier Aggregation Fundamentals

Carrier aggregation (CA) is the technique of combining multiple component carriers (CCs) to increase the effective bandwidth — and therefore throughput — available to a single UE. First introduced in LTE Release 10, CA has become the single most impactful feature for peak and average throughput gains in commercial networks.

In 5G NR, CA is specified in TS 38.300 Section 10 and the detailed band combination framework in TS 38.101-1 (FR1) and TS 38.101-2 (FR2). NR supports aggregation of up to 16 component carriers in the specification, though current commercial deployments typically use 2-4 CCs due to UE chipset and RF front-end constraints.

Each component carrier can have a bandwidth of up to 100 MHz in FR1 (sub-6 GHz) or 400 MHz in FR2 (mmWave), giving a theoretical maximum aggregated bandwidth of 1.6 GHz in FR1 or 6.4 GHz in FR2.

CA Types: Intra-Band vs Inter-Band

NR carrier aggregation is classified by the frequency relationship between component carriers.

Intra-Band Contiguous: The Simplest Case

When two CCs sit adjacent in the same band with no guard band gap, the UE can treat them as a single wideband carrier from an RF perspective. This is the most efficient form of CA because it requires minimal additional hardware — a single power amplifier and antenna path can serve both CCs.

In practice, intra-band contiguous CA is common in C-band deployments where an operator holds a large contiguous block. For example, T-Mobile US holds 100 MHz of contiguous n41 spectrum (2.5 GHz), which they aggregate as two 50 MHz CCs or deploy as a single 100 MHz carrier depending on the market.

Inter-Band CA: Coverage + Capacity

The most commercially valuable CA configuration combines a low-band coverage layer with a mid-band capacity layer. This gives the UE simultaneous access to wide-area coverage (for control plane robustness) and high bandwidth (for data throughput).

Common inter-band NR-CA combinations in commercial use:

EN-DC: The Bridge Between 4G and 5G

Before standalone 5G, EN-DC (E-UTRA NR Dual Connectivity) defined in TS 37.340 enables simultaneous LTE and NR connections. The UE maintains an anchor on the LTE eNB (master node) while adding an NR gNB (secondary node) for additional throughput.

EN-DC is technically dual connectivity rather than carrier aggregation — the key difference is that CA aggregates carriers within a single node's scheduler, while DC splits the bearer across two independent schedulers (LTE and NR). However, from a throughput perspective, the UE benefits from both connections simultaneously.

EN-DC Band Combinations

EN-DC band combinations are specified in TS 38.101-3. The notation uses the format DC_LTE-band_NR-band. Commercial examples:

• DC_1A-n78A: LTE B1 (2.1 GHz) + NR n78 (3.5 GHz) — widespread in Europe
• DC_3A-n78A: LTE B3 (1.8 GHz) + NR n78 (3.5 GHz) — common in Asia
• DC_66A-n77A: LTE B66 (AWS) + NR n77 (3.7 GHz) — North America
• DC_2A-n41A: LTE B2 (1.9 GHz) + NR n41 (2.5 GHz) — T-Mobile US

Throughput Calculation with Carrier Aggregation

The peak throughput for an NR CA configuration is calculated per TS 38.306 Section 4.1.2 using the formula applied to each component carrier, then summed:

`

Throughput_CC = v_layers x Q_m x f x R_max x (N_PRB x 12 / T_s) x (1 - OH)

`

Where:

• v_layers = number of MIMO layers
• Q_m = modulation order (8 for 256QAM)
• f = scaling factor (1.0)
• R_max = 948/1024
• N_PRB = number of PRBs for the channel BW and SCS
• T_s = OFDM symbol duration (including CP)
• OH = overhead (0.14 for DL FR1)

Worked Example: 2CC NR-CA on n78 + n1

CC1 — n78 (100 MHz, SCS 30 kHz): `

N_PRB = 273 PRBs

Symbols per slot = 14

Slots per subframe = 2 (30 kHz SCS)

Subcarriers per PRB = 12

Throughput_CC1 = 4 layers x 8 x 1.0 x (948/1024) x (273 x 12 x 14 x 2) / 0.001 x (1 - 0.14)

= 4 x 8 x 0.9258 x 128,105,600 / 0.001 x 0.86

= 4 x 8 x 0.9258 x 128,105,600,000 x 0.86

≈ 3,268 Mbps

` CC2 — n1 (20 MHz, SCS 15 kHz): `

N_PRB = 106 PRBs

Symbols per slot = 14

Slots per subframe = 1 (15 kHz SCS)

Throughput_CC2 = 4 layers x 8 x 0.9258 x (106 x 12 x 14 x 1) / 0.001 x 0.86

= 4 x 8 x 0.9258 x 17,808,000 / 0.001 x 0.86

≈ 454 Mbps

` Total peak throughput with 2CC CA: `

Total = 3,268 + 454 = 3,722 Mbps ≈ 3.7 Gbps

`

This represents a 14% throughput gain from adding the 20 MHz n1 carrier on top of the 100 MHz n78 primary. While the percentage gain seems modest per-carrier, the n1 CC provides crucial coverage continuity indoors and at cell edge where the n78 signal degrades.

Worked Example: EN-DC Throughput (LTE B3 + NR n78)

LTE CC — B3 (20 MHz, 2x2 MIMO, 256QAM): `

LTE peak DL = 2 layers x 8 x 0.9258 x 100 PRBs x 12 x 14 / 0.001 x (1 - 0.14)

≈ 195 Mbps (per TS 36.306 Category 19)

` NR CC — n78 (80 MHz, SCS 30 kHz, 4x4 MIMO): `

N_PRB = 217 PRBs

NR peak DL = 4 x 8 x 0.9258 x 217 x 12 x 14 x 2 / 0.001 x 0.86

≈ 2,598 Mbps

` Total EN-DC throughput: `

Total = 195 + 2,598 = 2,793 Mbps ≈ 2.8 Gbps

`

In practice, SK Telecom reports average EN-DC throughput of 800-1,200 Mbps in Seoul with DC_3A-n78A, versus 400-600 Mbps on NR n78 alone. The gap between theoretical peak and real-world average reflects multi-user scheduling, interference, propagation conditions, and UE capability limitations.

UE Capability and CA Support

Not all UEs support all CA combinations. UE CA capability is reported via RRC UE Capability Information (per TS 38.331 Section 5.6.1.4), which includes the specific band combinations, maximum number of CCs, MIMO layers per band, and supported bandwidth per CC.

The chipset determines the practical upper bound. A UE with the Snapdragon X75 can aggregate 300 MHz in sub-6 GHz (e.g., 3x100 MHz n77/n78) plus mmWave carriers for peak rates exceeding 10 Gbps in ideal conditions.

Scheduling and Resource Management Across CCs

When CA is active, the MAC scheduler at the gNB must jointly optimize resource allocation across all CCs. Per TS 38.321 Section 5.1, the UE receives grants on each CC independently but reports buffer status and channel quality across all CCs.

Key scheduling considerations:

Cross-carrier scheduling: The gNB sends DCI on one CC (scheduling CC) to allocate resources on another CC (scheduled CC). This reduces PDCCH overhead on capacity CCs. SCell activation/deactivation: Secondary cells can be dynamically activated and deactivated based on traffic demand. A timer-based mechanism (per TS 38.321) deactivates idle SCells to save UE power. Apple iPhone 15 Pro implements aggressive SCell deactivation, reducing CA-related battery drain by approximately 18% compared to always-on CA. Load balancing: The scheduler distributes traffic across CCs based on channel quality, congestion, and service requirements. URLLC traffic may be directed to the most reliable CC while eMBB traffic is spread across all available CCs.

Future of CA: Release 18+ Enhancements

3GPP Release 18 (5G-Advanced) introduces several CA enhancements under the work item "Further NR-CA Enhancements":

• Cross-carrier HARQ-ACK codebook: Reduces UCI overhead when operating with many CCs
• Multi-carrier scheduling DCI: A single DCI can schedule multiple CCs simultaneously per TS 38.212 Rel-18, reducing control overhead by up to 40%
• SCell dormancy enhancement: Finer-grained sleep states for secondary cells, improving UE energy efficiency
• CA with network energy saving: Coordinated CC sleep aligned with traffic patterns

Release 19 studies further expand CA to support up to 8 CCs in practical UE implementations and improve CA operation with reduced-capability (RedCap) devices for IoT use cases.

Key Takeaway: Carrier aggregation is the primary tool for maximizing 5G throughput. Understanding band combinations, the distinction between NR-CA and EN-DC, and the throughput calculation framework per TS 38.306 is essential for network planning, device evaluation, and performance benchmarking.