General statements: This document describes and summarizes only 5G communications

technologies and technical principles of Huawei products, and does not constitute any offers
or commitments. For detailed information about Huawei products or features, see the
product or feature documentation delivered with the software. Reproduction of this
document is prohibited.
1 Purpose
This document answers the following questions:
 What is carrier aggregation (CA)?

 Why is CA needed?

 How does secondary cell (SCell) management work?

 How do SCells and primary cells (PCells) work together to transmit and receive data?

 What is the relationship between CA and dual connectivity (DC)?

2 Overview
Enhanced Mobile Broadband (eMBB) is a key 5G application scenario. Its typical use cases,
as shown in Figure 2-1 , require extraordinarily high data rates:
 10 Gbit/s peak rate in the uplink

 20 Gbit/s peak rate in the downlink

Figure 2-1 Typical eMBB use cases

Peak rates are closely related to bandwidths available to UEs. If data is compared to
vehicles on highways, the cell bandwidth can be regarded as the number of lanes on a
highway. Under the same conditions, the more lanes, the higher the maximum rate that can
be reached across the highway.
In 5G, the maximum cell bandwidth can reach 100 MHz in low frequency bands and 400
MHz in high frequency bands. Take 100 MHz as an example. With this bandwidth, the
downlink peak rate of a UE can reach only 1.6+ Gbit/s (given a 30 kHz subcarrier spacing,
4-layer transmission, 256QAM modulation scheme, and 4:1 slot configuration). Such a peak
rate lags behind the requirements of eMBB use cases.
One option to meet the peak rate requirements of eMBB is to increase cell bandwidths.
However, the maximum bandwidth of a single cell can no longer be extended. To address
this requirement, the bandwidths of multiple cells can be aggregated for a single UE, as
shown in Figure 2-2 .
Figure 2-2 Example of aggregating bandwidths

This gives rise to CA. CA aggregates multiple component carriers (CCs) for a single UE so
that the UE can enjoy a bandwidth that equals the total bandwidth of the CCs. As a result,
the peak rate of the UE can increase almost proportionally.
Figure 2-3 presents an example of downlink peak rates.

Figure 2-3 Example of downlink peak rates

3 Basic Concepts of CA
3CC aggregation, as shown in Figure 3-1 , is one of the common applications of CA.

Figure 3-1 3CC aggregation

With 3CC aggregation, a UE is served by three carriers at the same time. Of these carriers,
one is the primary component carrier (PCC) and the other two are secondary component
carriers (SCCs). The cell on the PCC is referred to as the PCell, and the cell on each SCC is
referred as an SCell.
4 CA Specifications
From what is introduced above, it can be inferred that the core logic of CA is to utilize
SCCs to supplement bandwidths for the PCC and assist the PCC in data transmission and
reception. From this point of view, the specifications of CA lie in two aspects:
1. How many CCs can be aggregated at most?

2. What characteristics must CCs have before they can be aggregated?

Question 1: How many CCs can be aggregated at most?

Current 3GPP specifications stipulate that up to 16 CCs can be aggregated for NR CA and
the maximum aggregated bandwidth can reach 6.4 GHz (16 x 400 MHz), as shown in Figure
4-1.
Figure 4-1 16CC aggregation

The maximum number of CCs that can be aggregated may change with future protocol
updates. A similar case can be found with LTE. In 3GPP Release 10, a maximum of 5 CCs
could be aggregated to reach a maximum aggregated bandwidth of 100 MHz. Later in 3GPP
Release 13, this maximum number of CCs was extended to 32 and the maximum aggregated
bandwidth to 640 MHz (32 x 20 MHz).
Question 2: What characteristics must CCs have before they can be aggregated?
CCs must meet certain requirements before they can be aggregated. These requirements
mainly concern frequency bands and bandwidths.
Then, which band+bandwidth combinations are compatible with CA? Before getting to that
point, you need to get an overview of CA classification.
Based on whether CCs belong to the same frequency band and whether they are contiguous
in the frequency domain, CA can be categorized as follows:
1. Intra-band contiguous CA: CCs belong to the same frequency band and are contiguous in the frequency domain.

2. Intra-band non-contiguous CA: CCs belong to the same frequency band and are non-contiguous in the frequency domain.

3. Inter-band CA: CCs belong to different frequency bands. (Based on this, the CCs are basically non-contiguous in the frequency domain.)

Figure 4-2 uses 2CC aggregation as an example to illustrate intra-band contiguous CA,
intra-band non-contiguous CA, and inter-band CA.
Figure 4-2 Classification of CA
Technically speaking, intra-band contiguous CA is the easiest to implement. However, intra-
band non-contiguous CA and inter-band CA remain viable options because spectrum
resources owned by an operator are usually fragmented and non-contiguous. These two
types of CA fully utilize resources in each frequency band to provide higher peak rates for
UEs.
The preceding three categories of CA differ in their requirements for band+bandwidth
combinations. For details, see section 5.5A "Configurations for CA" in 3GPP TS 38.101-1
V16.6.0 and 3GPP TS 38.101-2 V16.6.0. Vendors can also customize the band+bandwidth
combinations based on the frequency bands and bandwidths supported by UEs.
In the following statements, Table in Protocols and the band+bandwidth combinations for various CA types are based on 3GPP TS 38.101-1 V16.6.0 (2020-12) and 3GPP TS 38.101-2
V16.6.0 (2020-12). The information is subject to changes in 3GPP specifications. For the latest information, see the updated 3GPP specifications.

Table 4-1 provides an overview of tables in protocols for various CA types.

Table 4-1 Tables in protocols for various CA types

CA Type Table in Protocols

Intra-band contiguous CA in Table 5.5A.1-1 in 3GPP TS

low frequency bands 38.101-1

Intra-band non-contiguous CA Table 5.5A.2-1 in 3GPP TS

in low frequency bands 38.101-1

Inter-band CA between low Table 5.5A.3.1-1 (2CC) and

frequency bands Table 5.5A.3.2-1 (3CC) in 3GPP
TS 38.101-1

Intra-band contiguous CA in Table 5.5A.1-1 in 3GPP TS

high frequency bands 38.101-2

Intra-band non-contiguous CA Table 5.5A.2-1 (with a single CA

in high frequency bands bandwidth class) and Table
5.5A.2-2 (with different CA
bandwidth classes) in 3GPP TS
38.101-2
Intra-Band Contiguous CA in Low Frequency Bands

Table 4-2 lists the band+bandwidth combinations supported for intra-band contiguous CA in low frequency bands.

Table 4-2 Band+bandwidth combinations for intra-band contiguous CA in low frequency bands

NR Uplink Chan Chan Chan Chan Chan Maxi Band

CA CA nel nel nel nel nel mum width
config configu band band band band band aggre combi
uratio rations width width width width width gated nation
n s for s for s for s for s for band set
carrie carrie carrie carrie carrie widt
r r r r r h
(MHz (MHz (MHz (MHz (MHz (MH
) ) ) ) ) z)

CA_n1 - 10 10, 15 40 0
NR Uplink Chan Chan Chan Chan Chan Maxi Band
CA CA nel nel nel nel nel mum width
config configu band band band band band aggre combi
uratio rations width width width width width gated nation
n s for s for s for s for s for band set
carrie carrie carrie carrie carrie widt
r r r r r h
(MHz (MHz (MHz (MHz (MHz (MH
) ) ) ) ) z)

B 15 15, 20

20 20

CA_n7 CA_n7 10, 10, 50 0

B B 15, 20 15,
20,
30, 40

CA_n4 - 20 80 100 0
0B
50 50

CA_n4 CA_n4 10, 10, 100 0

1B 1B 20, 20,
30, 30,
40, 50 40, 50

CA_n4 CA_n4 40 80, 180 0

1C 1C 100

50, 60,
60, 80 80,
100

10, 15, 190 1

15, 20,
20, 40,
40, 50,
50, 60,
60, 80,
80, 90 90,
100

CA_n4 - 20, 20, 40 100 0

6B 40, 60
NR Uplink Chan Chan Chan Chan Chan Maxi Band
CA CA nel nel nel nel nel mum width
config configu band band band band band aggre combi
uratio rations width width width width width gated nation
n s for s for s for s for s for band set
carrie carrie carrie carrie carrie widt
r r r r r h
(MHz (MHz (MHz (MHz (MHz (MH
) ) ) ) ) z)

CA_n4 - 60, 80 60, 80 160 0

6C

CA_n4 - 60, 80 80 80 240 0

6D

CA_n4 - 80 80 80 80 320 0
6E

CA_n4 - 20, 20, 40 20, 40 140 0

6M 40, 60

CA_n4 - 20, 20, 40 20, 40 20, 40 200 0

6N 40, 80

CA_n4 - 20, 60 20, 40 20, 40 20, 40 20, 40 220 0

6O

CA_n4 CA_n4 5, 10 10, 40 0

8B 8B 15, 20

15, 20 5, 10,
15, 20

- 10 50, 100 1
60,
80, 90

15, 20 40,
50,
60, 80

40 40,
50, 60

CA_n4 - 10 100 140 0

8C
15 90,
100
NR Uplink Chan Chan Chan Chan Chan Maxi Band
CA CA nel nel nel nel nel mum width
config configu band band band band band aggre combi
uratio rations width width width width width gated nation
n s for s for s for s for s for band set
carrie carrie carrie carrie carrie widt
r r r r r h
(MHz (MHz (MHz (MHz (MHz (MH
) ) ) ) ) z)

20 90,
100

40 80,
90,
100

CA_n6 - 51 20, 40 50 0
6B
10 15,
20, 40

15 10,
15, 20

20 5 1,
10, 15

40 5 1, 10

CA_n7 - 5 20 25 0
1B
10 15

15 10

20 5

10 20 35 1

15 15, 20

20 10, 15

CA_n7 CA_n7 50 60, 200 0

7C 7C 80,
100

60 60,
80,
NR Uplink Chan Chan Chan Chan Chan Maxi Band
CA CA nel nel nel nel nel mum width
config configu band band band band band aggre combi
uratio rations width width width width width gated nation
n s for s for s for s for s for band set
carrie carrie carrie carrie carrie widt
r r r r r h
(MHz (MHz (MHz (MHz (MHz (MH
) ) ) ) ) z)

100

80 80,
100

100 100

10, 10, 200 1

15, 15,
20, 20,
25, 25,
30, 30,
40, 40,
50, 50,
60, 60,
70, 70,
80, 80,
90, 90,
100 100

CA_n7 - 100 100 100 300 0

7D

CA_n7 - 20 50 70 0
8B

CA_n7 CA_n7 50 60, 200 0

8C 8C 80,
100

60 60,
80,
100

80 80,
100

100 100
NR Uplink Chan Chan Chan Chan Chan Maxi Band
CA CA nel nel nel nel nel mum width
config configu band band band band band aggre combi
uratio rations width width width width width gated nation
n s for s for s for s for s for band set
carrie carrie carrie carrie carrie widt
r r r r r h
(MHz (MHz (MHz (MHz (MHz (MH
) ) ) ) ) z)

10, 10, 200 1

15, 15,
20, 20,
25, 25,
30, 30,
40, 40,
50, 50,
60, 60,
70, 70,
80, 80,
90, 90,
100 100

CA_n7 - 100 100 100 300 0

8D

CA_n7 CA_n7 50 60, 200 0

9C 9C 80,
100

60 60,
80,
100

80 80,
100

100 100

CA_n7 - 100 100 100 300 0

9D

NOTE 1: 5 MHz is not applicable for 30/60 kHz SCS.

 NR CA configuration: indicates the applicable frequency bands and the number of CCs for CA. For example, configuration CA_n1B indicates that contiguous CA works in band
n1 and that 2CC aggregation is supported. In this expression, B is a CA bandwidth class. This value indicates contiguous 2CC. Figure 4-3 provides the definitions of CA
bandwidth classes. For details, see Table 5.3A.5-1 "NR CA bandwidth classes" in 3GPP TS 38.101-1.
 Figure 4-3 NR CA bandwidth classes for low frequency bands

Class A means 1CC, in which case CA is not supported.

 Uplink CA configurations: indicates whether uplink CA is supported. If any information is present in a cell under this column, uplink CA with that specified configuration is
supported. If no information is present, uplink CA is not supported. For example, the Uplink CA configurations column is empty for configuration CA_n1B, indicating that CA
does not work in the uplink in band n1. For configuration CA_n7B, the value of Uplink CA configurations is CA_n7B, indicating that 2CC aggregation works in the uplink in
band n7.

 Channel bandwidths for carrier (MHz): indicates the bandwidth of each CC. The third to seventh columns list the channel bandwidths for CCs. The values in these columns
can be randomly combined. For example, in band n1, all the following bandwidth combinations are allowed for CA: 10 MHz + 10 MHz, 10 MHz + 15 MHz, 15 MHz + 15 MHz, 15
MHz + 20 MHz, and 20 MHz + 20 MHz.

 Maximum aggregated bandwidth (MHz): equals the maximum sum of the individual CC bandwidths.

 Bandwidth combination set: identifies each set of bandwidth combinations for a single band combination. It is introduced to facilitate UE capability identification, given that
bandwidth combinations supported by 3GPP specifications are continuously changing. If a UE reports "0", it indicates that only bandwidth combination set 0 for the specified
band combination is supported.

Intra-Band Non-Contiguous CA in Low Frequency Bands

Table 4-3 lists the band+bandwidth combinations supported for intra-band non-contiguous CA in low frequency bands.

Table 4-3 Band+bandwidth combinations for intra-band non-contiguous CA in low frequency bands

NR CA Uplink Chan Chan Chan Chan Maxi Bandw

Configu Configu nel nel nel nel mum idth
ration rations band band band band Aggre combi
widths widths widths widths gated nation
for for for for band set
carrie carrie carrie carrie width
r r r r (MHz)
(MHz) (MHz) (MHz) (MHz)

CA_n3( - 5, 10, 5, 10, 40 0

2A) 15, 20 15, 20

CA_n7( - 5, 10, 5, 10, 40 0

2A) 15, 20 15, 20

CA_n25 - 5, 10, 5, 10, 40 0

NR CA Uplink Chan Chan Chan Chan Maxi Bandw
Configu Configu nel nel nel nel mum idth
ration rations band band band band Aggre combi
widths widths widths widths gated nation
for for for for band set
carrie carrie carrie carrie width
r r r r (MHz)
(MHz) (MHz) (MHz) (MHz)

(2A) 15, 20 15, 20

CA_n41 CA_n41 40, 50, 40, 50, 180 0

(2A) (2A) 60, 80, 60, 80,
100 100

10, 15, 10, 15, 190 1

20, 40, 20, 40,
50, 60, 50, 60,
80, 90, 80, 90,
100 100

CA_n48 10, 15, 10, 15, 1402 0

(2A) 20, 40, 20, 40,
50, 60, 50, 60,
80, 90, 80, 90,
100 100

CA_n48 - 10, 15, 10, 15, 10, 15, 1402 0

(3A) 20, 40, 20, 40, 20, 40,
50, 60, 50, 60, 50, 60,
80, 90, 80, 90, 80, 90,
100 100 100

CA_n48 - 10, 15, 10, 15, 10, 15, 10, 15, 1352 0
(4A) 20, 40, 20, 40, 20, 40, 20, 40,
50, 60, 50, 60, 50, 60, 50, 60,
80, 90, 80, 90, 80, 90, 80, 90,
100 100 100 100

CA_n66 - 5, 10, 5, 10, 60 0

(2A) 15, 20, 15, 20,
40 40

CA_n77 CA_n77 20, 40, 20, 40, 200 0

(2A) (2A) 80, 80,
100 100

CA_n78 CA_n78 10, 20, 10, 20, 200 0

 NR CA Uplink Chan Chan Chan Chan Maxi Bandw
Configu Configu nel nel nel nel mum idth
ration rations band band band band Aggre combi
widths widths widths widths gated nation
for for for for band set
carrie carrie carrie carrie width
r r r r (MHz)
(MHz) (MHz) (MHz) (MHz)

(2A) (2A) 40, 50, 40, 50,

60, 80, 60, 80,
90, 90,
100 100

10, 20, 10, 20, 200 1

25, 30, 25, 30,
40, 50, 40, 50,
60, 80, 60, 80,
90, 90,
100 100

10, 20, 10, 20, 200 2

25, 30, 25, 30,
40, 50, 40, 50,
60, 70, 60, 70,
80, 90, 80, 90,
100 100

NOTE 1: Void.
NOTE 2: Parameter value accounts for both, the maximum
frequency range of band n48 (150 MHz), and the minimum
frequency gaps in between NR non-contiguous component
carriers.

 NR CA Configuration: indicates the applicable frequency bands and the number of CCs for CA. For example, configuration CA_n3(2A) indicates that non-contiguous CA works
in band n3 and that 2CC aggregation is supported. In this expression, 2A means two non-contiguous CCs. It cannot be replaced with B, which signifies two contiguous CCs.

 The definitions of the other columns are the same as those for contiguous CA.

Inter-Band CA Between Low Frequency Bands

Figure 4-4 presents only some rows of Table 5.5A.3.1-1 (2CC) in 3GPP TS 38.101-1.
Figure 4-4 Band+bandwidth combinations for inter-band CA between low frequency bands

 NR CA configuration: indicates the applicable frequency bands and the number of CCs for CA. For example, configuration CA_n1A_n3A indicates that one carrier (with
bandwidth class A) in band n1 can be aggregated with one carrier (with bandwidth class A) in band n3.

 Uplink CA configuration: indicates whether uplink CA is supported. If any information is present in a cell under this column, uplink CA with that specified configuration is
supported. If no information is present, uplink CA is not supported.

 NR Band: indicates the bands for CA. For example, the value n1 indicates that band n1 supports inter-band CA.

 The definitions of the other columns are the same as those for intra-band contiguous CA in low frequency bands.

Intra-Band Contiguous CA in High Frequency Bands

Figure 4-5 presents only some rows of Table 5.5A.1-1 in 3GPP TS 38.101-2.

Figure 4-5 Band+bandwidth combinations for intra-band contiguous CA in high frequency bands

 The definitions of the NR CA configuration, Uplink CA configurations, BW Channel (MHz), Maximum aggregated BW (MHz), and BCS columns are the same as those of the
counterpart columns for intra-band CA in low frequency bands. (BCS is short for bandwidth combination set.)

 The definitions of CA bandwidth classes for high frequency bands are available in Figure 4-6 .
 Figure 4-6 NR CA bandwidth classes for high frequency bands

Intra-Band Non-Contiguous CA in High Frequency Bands

The concept of sub-blocks has been introduced for intra-band non-contiguous CA in high frequency bands. A sub-block consists of one or more contiguous carriers. This type of CA is
equivalent to contiguous CA within each sub-block plus aggregation of the sub-blocks.

For intra-band non-contiguous CA in high frequency bands, the CA bandwidth classes of sub-blocks can be either identical or varied.

 Figure 4-7 presents only some rows of Table 5.5A.2-1 in 3GPP TS 38.101-2, with the CA bandwidth classes of sub-blocks being identical.

Figure 4-7 CA bandwidth classes of sub-blocks being identical

 The definitions of the NR configuration and Uplink CA configurations columns are the same as those in the preceding tables.

 The definition of sub-blocks has been provided previously. As shown in the table, the CA bandwidth classes of the sub-blocks are identical.

 indicates the maximum bandwidth that can be reached after the sub-blocks are aggregated.

 The definition of the BCS column has been provided previously.

 Figure 4-8 presents only some rows of Table 5.5A.2-2 in 3GPP TS 38.101-2, with the CA bandwidth classes of sub-blocks being varied.
 Figure 4-8 CA bandwidth classes of sub-blocks being varied

As shown in the table, the CA bandwidth classes of the sub-blocks are varied. For example, configuration CA_n260(2A-D) in the second entry indicates aggregation of three
sub-blocks: CA_n260A, CA_n260A, and CA_n260D.

FDD+TDD CA

As described above, CA can be classified into intra-band CA and inter-band CA depending on whether CCs belong to the same frequency band. FDD+TDD CA must be a type of inter-band
CA, because FDD and TDD CCs belong to different frequency bands.

With FDD+TDD CA, a CA UE can perform uplink and downlink data transmission on both FDD and TDD carriers. Operators can designate an FDD or TDD carrier as the PCC, if required, as
shown in Figure 4-9 .

Figure 4-9 FDD+TDD CA

What are the major difficulties in FDD+TDD CA? The PCell and SCells in FDD+TDD CA work in different duplex modes and therefore use different HARQ feedback timing. Without special
handling of HARQ feedback timing, FDD+TDD CA will come up with out-of-order HARQ feedback, which in turn may cause failures to normally transmit or receive uplink control information in
the PCell and SCells. How to handle HARQ feedback timing is not specified in protocols and therefore varies depending on vendors.

5 What Kind of Information UEs Need to Report for CA

The information in 4 CA Specifications is about CA capabilities from the perspective of
gNodeBs. CA capabilities described below are from the perspective of UEs, as UE
capabilities are important to the implementation of CA. The CA capabilities that a UE can
actually acquire are determined by both the CA capabilities of the UE and those of the
gNodeB that the UE is connected to.
The CA capabilities of UEs mainly refer to the band combinations supported by the UEs.
After a gNodeB sends a UECapabilityEnquiry message to query UE capabilities, the UE
responds with a UECapabilityInformation message, which conveys the CA capabilities of
the UE, as shown in Figure 5-1 .
Figure 5-1 CA capability reporting procedure

A major information element (IE) in this context is BandCombination. This IE contains the
band combinations supported by the UE. Figure 5-2 explains this IE.
Figure 5-2 Explanation of BandCombination

The gNodeB performs CA for the UE based on the reported UE capabilities. In the example
provided in Figure 5-2 , the UE supports intra-band CA in band n78. Then, the gNodeB will
search for a cell in band n78 and configure it as an SCell for the UE.
6 How SCell Management Works
The philosophy of CA is to utilize SCells to extend the available bandwidths for UEs so that
the peak rates of the UEs can increase. This necessitates two key techniques for CA:
 How to find SCells for UEs using CA and, after they are found, how to manage them

 How PCells and SCells work together to transmit and receive data

These are the essential aspects of the CA technology. Now, let's have a look at how they
work.
Management typically refers to addition, removal, modification, and query. In addition to
these activities, SCell management includes SCell activation and deactivation. Figure 6-
1 illustrates the relationships between SCell states.

In this context, SCell configuration refers to SCell addition.

Figure 6-1 SCell state transitions

CA takes effect for individual UEs, so the SCell management activities in this figure are
specific to an individual UE.
1. SCell configuration: New SCells are configured for the UE.

2. SCell activation: Only activated SCells can transmit and receive data. After being configured, SCells are not activated until certain conditions are met.

3. SCell deactivation: Deactivated SCells cannot transmit or receive data.

4. SCell change: SCells are changed to cells with better signal quality for the UE.

5. SCell removal: SCells whose signal quality drops below a specified threshold are removed.

SCell activation and deactivation are introduced to reduce costs, including those of the UE
and SCells.
When an SCell is active, the UE and SCell need to exchange control information. This
consumes not only UE power, but also the radio resources of the SCell. When the data
traffic of the UE is not heavy, the PCell itself can handle all the services of the UE, without
the need for extra resources. Therefore, activation and deactivation are introduced.
 SCell configuration

A CA frequency group and frequencies in the group are manually configured on the
gNodeB. The cells on these frequencies are candidate SCells. After an RRC
 connection is set up between a UE and its PCell, the PCell determines the SCells to
be added for the UE, based on the capabilities of the UE and candidate SCells. The
PCell then sends the information about the selected SCells to the UE through an RRC
reconfiguration message, and the UE adds SCells based on the SCell information, as
illustrated in Figure 6-2 . There are two methods available for the UE to add SCells:
blind and measurement-based SCell configuration. When blind SCell configuration is
used, the UE configures SCells based on the SCell information delivered by the
gNodeB. When measurement-based SCell configuration is used, the UE takes the
signal quality of cells into account. A cell can be configured as an SCell only when
its signal quality meets certain conditions.
Figure 6-2 SCell configuration

 SCell activation and deactivation

SCells are inactive upon being added. The gNodeB determines when to switch the
SCells to the active state. Generally, the gNodeB determines whether to activate
SCells based on the traffic volume of the UE. When the traffic volume of the UE is
heavy, the gNodeB activates SCells.
Figure 6-3 shows SCell activation.

Figure 6-3 SCell activation

When does an active SCell switch to the inactive state? This can be determined by
the gNodeB or the UE.
 The gNodeB deactivates SCells when data traffic is light.
  The UE sets a deactivation timer for each SCell. The timer is started for an SCell upon SCell activation. If no control or service data is detected within the timer
length, the SCell is automatically deactivated. (The UE receives the timer length during SCell configuration.)

 SCell change

SCell change is a procedure for changing an SCell for a UE. After an SCell is
configured for a UE, the UE measures the signal quality of cells on the same
frequency as the SCell. If one of the cells has a better signal quality than the current
SCell, an SCell change takes place by removing the current SCell and then
configuring the better cell as an SCell for that UE.
In this procedure, event A6 is used. The related measurement configurations are
included in an RRC reconfiguration message.
Figure 6-4 shows SCell change.

Figure 6-4 SCell change

 SCell removal

SCell removal is a procedure for removing an SCell from the SCell list of a UE.
After an SCell is configured for a UE, the UE periodically measures the signal
quality of the SCell. If the signal quality is lower than a specified threshold, the
SCell is removed.
In this procedure, event A2 is used. The related measurement configurations are
included in an RRC reconfiguration message.
Figure 6-5 shows SCell removal.
 Figure 6-5 SCell removal

The preceding descriptions of "SCell change" are based on the assumption that the PCell
remains unchanged. When the PCell changes (by handover), an SCell change is
automatically triggered. After a successful handover, the target PCell automatically searches
for surrounding cells that meet certain conditions, and then it configures new SCells for the
UE with the original SCells removed. If no SCell meets the conditions, the original SCells
are retained for the UE.
7 How SCells and PCells Work Together to Transmit and Receive
Data
The PCell and SCells of a UE work together to transmit and receive data. How do they
collaborate with each other in this process? According to the downlink protocol stack with
CA as shown in Figure 7-1 , CCs are aggregated mainly at the Media Access Control (MAC)
layer on the air interface. To learn how the PCell and SCells collaborate with each other,
you need to understand the implementation of these cells at the MAC layer.
Figure 7-1 Downlink protocol stack with CA

The major functions of the MAC layer are hybrid automatic repeat request (HARQ) and
scheduling.
 HARQ feedback

As shown in the figure, there is one independent HARQ entity per carrier. HARQ feedback
is transmitted over the PUCCH, which in CA scenarios is present only in the PCell. To be
specific, HARQ feedback of all serving cells is transmitted over the PUCCH in the PCell.
To carry the HARQ feedback of all serving cells, the PUCCH needs to use a format that can
accommodate more bits of information. In NR CA, the PUCCH uses format 3. Format 3
offers the largest payload capacity among the five PUCCH formats. It occupies 4 to 14
symbols, which can convey more than 2 bits of information. For details about the PUCCH,
see Uplink Physical Channels - Introduction to Control Channels and Signals in the
technology series.
 Scheduling

Without CA (and any other multi-carrier technologies such as dual connectivity [DC]), the
data of a UE can be transmitted by using only one carrier. That means all UE data is
scheduled and transmitted in the cell on the carrier. With CA, the data of a UE can be
transmitted over both the PCC and SCCs. In that case, how do the PCC and SCCs
collaborate for scheduling? This can be interpreted as two questions: Which data is
scheduled on the PCC and which data is scheduled on the SCCs? How can we ensure that
the data over the PCC and that over SCCs "synchronously" arrive at the UE or gNodeB?
First, which data is scheduled on the PCC and which data is scheduled on the SCCs? This
has not been defined in the protocols. Its implementation varies according to vendors. The
following are possible solutions:
 The signal quality of the PCC and SCCs determines how much data can be scheduled separately on the CCs in each scheduling period, as illustrated in Figure 7-2 . If the signal
quality of the PCC is better than that of an SCC, more data is scheduled on the PCC; otherwise, more data is scheduled on the SCC.

Figure 7-2 Data scheduling between the PCC and an SCC

 The service type of data determines whether to schedule the data on the PCC only or on both the PCC and SCCs. In the case of scheduling on both the PCC and SCCs, data
split is necessary. Data split involves an interaction between the PCC and SCCs, with this interaction inevitably causing delays. For services with stringent requirements on
delay, such as video services, data split can be replaced with scheduling only on the PCC.

Second, how can we ensure that the data over the PCC and that over SCCs "synchronously"
arrive at the UE or gNodeB? "Synchronously" herein does not mean being exactly at the
same time. It means that the time difference between data arrivals at the UE or gNodeB is
acceptable. There is no need to pay too much concern to this issue. A time difference
tolerable by the UE or gNodeB is generally at the 100 ms level. The time required for data
transmission over the air interface can meet requirements as long as the air interface quality
is not too poor. (In fact, even in non-CA scenarios, different packets of the same service
cannot arrive at the UE or gNodeB at the same time.)
8 What Is the Relationship Between CA and DC
When it comes to CA, another technology that needs to be mentioned is DC.
CA and DC are essentially the same in that a second, third, or even more cells are used to
serve a single UE in order to obtain more bandwidth resources for data transmission and
improve user experience. Why are there two technologies in parallel? What are the
differences between them?
First, in concepts:
1. DC is a mode of operation for RRC_CONNECTED UEs. When the network configures a master cell group (MCG) and a secondary cell group (SCG) for a UE in
RRC_CONNECTED mode, the UE enters the DC state. CA refers to aggregation of multiple carriers in the same gNodeB (known as intra-gNodeB CA) or between gNodeBs
(known as inter-gNodeB CA).

Figure 8-1 illustrates DC and CA.

 Figure 8-1 DC and CA

3GPP specifications have defined intra-gNodeB CA, but not inter-gNodeB CA. Vendors can offer inter-gNodeB CA functionalities beyond the scope of 3GPP specifications. Like
DC, inter-gNodeB CA refers to aggregation of multiple carriers between two gNodeBs. It can be intra-band CA (including contiguous and non-contiguous CA) or inter-band CA.
For inter-gNodeB CA, an eXn interface is used to transmit signaling and data between the gNodeBs.

2. The base stations engaged in DC can be of different RATs, but those engaged in CA must be of the same RAT. DC between base stations of different RATs is also known as
MR-DC. (Non-standalone [NSA] networking is a type of MR-DC. It is a kind of native DC.)

3. The layer at which data split occurs is PDCP for DC but RLC or MAC for CA. That means DC is implemented at a higher protocol layer. This is the major reason why DC and
CA are used in different scenarios.

Second, in terms of application scenarios:

1. DC is recommended when the delay on the bearer network (simply understood as the network between base stations) is relatively long. Data split for DC occurs at the PDCP
layer, which has a relatively loose requirement for delay. However, data split for CA occurs at the RLC or MAC layer, which is below the PDCP layer and has a relatively
stringent requirement for delay. Therefore, DC can be used when the delay of specific services (such as video services) may not meet the requirements on the bearer network.

2. In addition to the preceding difference, the selection between CA and DC in the industry is dependent on a UE's capabilities of CA and DC and also operator requirements. It is
believed that with the increasingly wide use of CA and DC in the future, the differences in application scenarios between CA and DC will be clearer.

As mentioned above, NSA is a type of native DC. Is NSA DC intended for LTE-to-NR
evolution or rate increase? The answer is derived from the development of DC.
The concept of DC was introduced as early as in LTE Release 12. CA has a relatively
stringent requirement for delay on the bearer network, which meant some operators were
unable to meet this requirement. DC was therefore proposed by 3GPP for its looser delay
requirements. However, DC did not witness its wide use in LTE, mainly due to limited UE
capabilities. In the DC architecture, the primary and secondary nodes work in parallel on the
user plane. Therefore, 2T2R UEs are required. In the LTE era, the commercial use of 2T2R
UEs was not implemented on a large scale, limiting the development of DC.
With NR, the situation is different. NSA networking is introduced to meet evolution
requirements. It is a kind of native DC. For NSA development, 2T2R UEs must be
available. Since the requirement for UEs is addressed, DC is being developed as it is a key
solution to lift UE rates.
Figure 8-2 illustrates the conceptual differences between NSA networking (MR-DC) and
NR CA.
Figure 8-2 Conceptual differences between NSA networking and NR CA

The primary base station is an eNodeB, also referred to as a master eNodeB (MeNB). The
secondary base station is a gNodeB, also referred to as a secondary gNodeB (SgNB). CA
can be performed separately within the eNodeB and within the gNodeB. The cells served by
the eNodeB are aggregated in CA to form the MCG, and the cells served by the gNodeB are
aggregated in CA to form the SCG. The PCC in the SCG is also called the primary
secondary component carrier (PSCC).
For more details about NSA networking, see Initial Access (NSA Networking) in the
technology series.
9 CA-related Protocols

Protocol Section Content

Number

38.300 5.4/6.7/7.7 CA implementation at

the physical layer and
MAC layer, as well as
CA-related RRC
implementation

38.331 4.4/5.3.5.5.8/5.3.5.5.9/5.6 SCell management

definitions, SCell
removal, SCell
Protocol Section Content
Number

configuration, SCell
change, and UE
capabilities

38.321 5.9/6.1.3.10 SCell

activation/deactivation
logic and MAC CEs

38.213 4.3 Timing for SCell

activation/deactivation

38.101 5.5 Band+bandwidth

combinations
10 References
1. 3GPP TS 38.321, NR; Medium Access Control (MAC) protocol specification (Release 16)

2. 3GPP TS 38.331, NR; Radio Resource Control (RRC) protocol specification (Release 16)

3. 3GPP TS 38.101-1, NR; User Equipment (UE) radio transmission and reception; Part 1: Range 1 Standalone (Release 16)

4. 3GPP TS 38.101-2, NR; User Equipment (UE) radio transmission and reception; Part 2: Range 2 Standalone (Release 16)

5. 3GPP TS 38.300, NR; NR and NG-RAN Overall Description; Stage 2 (Release 16)

6. 3GPP TS 38.213, NR; Physical layer procedures for control (Release 16)

7. Erik Dahlman, Stefan Parkvall, and Johan Sköld. 5G NR: The Next Generation Wireless Access Technology. Academic Press, 2018