1

Understanding the Heart of the 5G Air Interface: An Overview of

Physical Downlink Control Channel for 5G New Radio (NR)
Kazuki Takeda‡, Huilin Xu‡, Taehyoung Kim§, Karol Schober⊥, and Xingqin Lin#
‡
Qualcomm Inc., §Samsung Electronics, ⊥Nokia, #Ericsson Inc.
Emails: [email protected]; [email protected]; [email protected];
[email protected]; [email protected].

Abstract— New Radio (NR) is a new radio air interface grouped to form a physical resource block (PRB), which is used
developed by the 3rd Generation Partnership Project (3GPP) for as a unit in frequency-domain resource allocation for
the fifth generation (5G) mobile communications system. With channels/signals. Fourteen OFDM symbols for the given SCS
great flexibility, scalability, and efficiency, 5G is expected to
are grouped to form a slot (a time unit commonly recognized
address a wide range of use-cases including enhanced mobile
broadband (eMBB), ultra-reliable low-latency communications among sets of user equipment (UEs) for the given SCS), where
(URLLC), and massive machine type communications (mMTC). the slot duration is 2−𝜇 ms. One or more slots comprise a 1 ms
The physical downlink control channel (PDCCH) in NR carries subframe, and 10 subframes are grouped to form a radio frame
Downlink Control Information (DCI). Understanding how [5]. For carrier aggregation (CA), the SCS can be different
PDCCH operates is key to developing a good understanding of across serving cells, i.e., carriers.
how information is communicated over NR. This paper provides
(2) Flexible bandwidth: NR is designed such that individual
an overview of the 5G NR PDCCH by describing its physical layer
structure, monitoring mechanisms, beamforming operation, and UEs can use smaller bandwidths than the system bandwidth.
the carried information. We also share various design rationales The UE bandwidth within a carrier, configured by the base
that influence NR standardization. station (BS) as the number of contiguous PRBs with an
Index Terms—5G, New Radio, PDCCH, beamforming, mobile associated SCS is called a bandwidth part (BWP) [6]. Once the
communication BWP is activated, data and control channels are
received/transmitted within the BWP [7]. A UE can be
I. INTRODUCTION configured with up to four BWPs, where BWPs may have
HE 3rd Generation Partnership Project (3GPP) has
T developed the initial release of the fifth generation (5G)
mobile communications system as 3GPP Release 15 [1]. New
different SCSs and may be mutually overlapping or non-
overlapping in frequency. If more than one BWP is configured
for a UE, the BS may select which BWP is active at a given
Radio (NR) is a radio access technology newly introduced in time by means of downlink (DL) control. Therefore, the BS
3GPP Release 15. Though NR is not required to support may adjust dynamically the UE bandwidth according to the
backward compatibility with long-term evolution (LTE), amount/profile of data traffic for the UE. A reduction in the UE
ensuring harmonious LTE-NR coexistence has also been an bandwidth may reduce the UE power consumption.
important design consideration. From the physical layer perspective, the data and signaling
5G is expected to fulfill various requirements from a wide messages in NR are carried in the DL and uplink (UL) physical
range of industries/verticals so that many use-cases such as channels. Among these channels, the physical downlink control
enhanced mobile broadband (eMBB), ultra-reliable low-latency channel (PDCCH) plays a central role in, for example, DL
communications (URLLC), and massive machine type scheduling assignments, UL scheduling grants, and special
communications (mMTC), are supported [2-3]. In addition, 5G purposes such as slot format indication, preemption indication,
must support an extremely wide frequency range from below 1 and power control. Understanding how the PDCCH operates is
GHz to above 50 GHz, which includes the frequencies that have key to developing a good understanding of how information is
not been utilized for cellular networks to date, e.g., millimeter communicated over NR.
waves [3]. New scenarios, use cases, requirements, and NR PDCCH has been studied and discussed in [5][9][10].
frequencies have posed many new challenges. Performance study of wideband and narrowband precoding of
3GPP Release 15 NR is designed to address all these NR PDCCH and a design of PDCCH hashing function can be
challenges. Some of the NR characteristics are highlighted found in [5]. The complexity analysis of NR PDCCH is
below and further details can be found in [4]. presented in [9]. Brief overview of NR PDCCH is presented in
(1) Orthogonal frequency domain multiple access [10]. In this article, we present a comprehensive A to Z
(OFDMA) with flexible subcarrier spacing (SCS): NR overview of the final NR PDCCH design as well as we explain
adopts OFDMA as its multiple access scheme. SCS values of reasons that led to such design.This article presents the physical
15 × 2𝜇 kHz with μ = 0, 1, 2, 3, or 4 are supported [5]. The layer design of the NR PDCCH in Section II. Section III
system selects SCS values based on carrier frequency, use-case, describes how a UE receives the NR PDCCH. In Section IV,
scenario, and/or requirements, and the SCS is configured using beamforming operation of the NR PDCCH is explained.
higher-layer signaling. Once the SCS is configured, frame Section V provides details of the information carried by NR
structure is identified. Twelve subcarriers for a given SCS are PDCCH. Section VI summarizes the article.
 2

CRC RNTI Polar Sub-block

DCI attachment masking
Interleaving
coding interleaving

Rate QPSK
Scrambling RE mapping Precoding
matching modulation

CCE CCE-to-REG
indexing mapping

Fig. 1 Procedure for generating PDCCH from DCI.

aggregation level (AL). With QPSK modulation, a CCE

II. PDCCH PHYSICAL LAYER STRUCTURE contains 54 payload REs and therefore can carry 108 bits. This
requires the output size of the rate matching block to be 𝐿 ⋅ 108,
A. Channel Coding and Downlink Control Information (DCI)
Construction where 𝐿 is the associated AL. Based on the channel
environment and available resources, the BS can adaptively
The information carried by the PDCCH is referred to as DCI.
choose a proper AL for a DCI to adjust the code rate.
DCI contains the scheduling information for the UL or DL data
channels and other control information for one UE or a group B. Control Resource Sets
of UEs. Further details regarding DCI are provided in Section A DCI with AL L is mapped to physical resources in a given
V. DCI undergoes a sequence of processing steps to constitute BWP, where necessary parameters such as frequency and time
the PDCCH payload. domain resources, and scrambling sequence identity for the
The procedure for generating a PDCCH is illustrated in Fig. DMRS for the PDCCH are configured to a UE by means of
1. If the size of the DCI format is less than 12 bits, a few zero control resource set (CORESET). In this subsection, we briefly
padding bits will be appended until the payload size equals 12 introduce the CORESET, and then in Subsection II-C, we
bits [11]. For the DCI payload bits, a 24-bit cyclic redundancy overview how the DCI with AL L is mapped in the CORESET
check (CRC) is calculated and appended to the payload. The as a PDCCH.
CRC allows the UE to detect the presence of errors in the A UE may be configured with up to three CORESETs on
decoded DCI payload bits. After the CRC is attached, the last each of up to four BWPs on a serving cell. Therefore, a UE may
16 CRC bits are masked with the corresponding identifier, be configured with up to 12 CORESETs on a serving cell in
referred to as a radio network temporary identifier (RNTI). total, where each CORESET has an index of 0-11. In general,
Using the RNTI mask, the UE can detect the DCI for its unicast CORESETs are configured in units of six PRBs on a six PRB
data and distinguish sets of DCI with different purposes that frequency grid (starting from a reference point referred to as
have the same payload size. The CRC attached bits are then point A) and one, two, or three consecutive OFDM symbols in
interleaved to distribute the CRC bits among the information the time domain. There is a special CORESET with index 0
bits. The interleaver supports a maximum input size of 164 bits. (CORESET 0), which is configured using a four-bit information
This means that DCI without CRC can have at most 140 of element in the master information block (MIB) with respect to
payload bits. The bits are then encoded by the Polar encoder to the cell-defining synchronization signal and physical broadcast
protect the DCI against errors during transmission. The encoder channel (PBCH) block (SSB). CORESET 0 can be acquired
output is processed using a sub-block interleaver and then rate even before higher-layer configurations are provided, i.e.,
matched to fit the allocated payload resource elements (REs) of before additional system information or dedicated
the DCI. configuration is provided, while its configuration is restricted
The payload bits of each DCI are separately scrambled by a to a limited number of combinations of parameters compared to
scrambling sequence generated from the length-31 Gold CORESETs with indices other than 0. Moreover, CORESET 0
sequence. The scrambling sequence is initialized by the may not be aligned with a six PRB frequency grid for other
physical layer cell identity of the cell or by a UE specific CORESETs, since the frequency-domain resource of
scrambling identity and a UE specific cell RNTI (C-RNTI). CORESET 0 is determined relatively with the SSB as illustrated
After the scrambled DCI bit sequence is Quadrature Phase Shift in Fig. 2 as an example, in which CORESET 0 is illustrated in
Keying (QPSK) modulated, the complex-valued modulation red and CORESET 1 is illustrated in yellow on a BWP of 48
symbols are mapped to physical resources in units referred to PRBs.
as control channel elements (CCEs). Each CCE consists of six CORESETs are active only when their associated BWP is
resource element groups (REGs), where a REG is defined as active, with the exception of CORESET 0, which is associated
one PRB in one OFDM symbol which contains nine REs for the with the initial BWP (BWP with index 0). However, CORESET
PDCCH payload and three demodulation reference signal 0 can be monitored in another BWP when its numerology, i.e.,
(DMRS) REs. For each DCI, 1, 2, 4, 8, or 16 CCEs can be SCS 𝜇 and cyclic prefix, matches the numerology of the non-
allocated, where the number of CCEs for a DCI is denoted as
 3

Point A

BWP (48 PRBs)

Frequency
PRB CORESET 0 (24 PRBs)

Six PRB frequency grid CORESET 1 (36 PRBs on six PRB frequency grid)
Fig. 2 Example of CORESET configurations on BWP.

initial BWP, and the PRBs of CORESET 0 are confined within mapping, all CCEs for a DCI with AL L are mapped in
the PRBs of the non-initial BWP. consecutive REG bundles of the CORESET.
C. Mapping of PDCCH to CORESET Once the REGs corresponding to a PDCCH are determined,
the modulated symbols of the PDCCH are mapped to the REs
A DCI of AL 𝐿 comprises 𝐿 continuously numbered CCEs, of the determined REGs in the frequency domain first and the
and the CCEs are mapped on a number of REGs in a CORESET. time domain second, i.e. in increasing order of the RE index and
NR supports distributed and localized resource allocation for symbol index, respectively.
a DCI in a CORESET. This is done by configuring interleaved NR supports both wideband and narrowband precoding for
or non-interleaved CCE-to-REG mapping for each CORESET. the PDCCH. In wideband precoding, PDCCH DMRSs are
For interleaved CCE-to-REG mapping, REG bundles transmitted in all contiguous REGs of a CORESET carrying the
constituting the CCEs for a PDCCH are distributed in the PDCCH using the same precoder. On the other hand, in the
frequency domain in units of REG bundles as illustrated in Fig. narrowband precoding, PDCCH DMRSs are transmitted only
3. A REG bundle is a set of indivisible resources consisting of in the REG bundles actually used for the PDCCH transmission,
neighboring REGs. A REG bundle spans across all OFDM and precoding is constant only within the REG bundle. The
symbols for the given CORESET. Therefore, interleaved CCE- former can maximize the frequency domain processing gain
to-REG mapping can enable both a time domain processing and the latter can take advantage of frequency dependent
gain and frequency domain diversity. Interleaved CCE-to-REG beamforming gain. Since the same precoding is applied to
mapping can be visualized as a process for which REG bundle payload and the corresponding DMRS, precoding is transparent
indices are continuously filled in an array row-wise first and to the UE.
then read out column-wise. This process is often called block
interleaving. By this means, adjacent CCEs for a PDCCH are III. PDCCH MONITORING
broken down into scattered REG bundles in the frequency In this section, we discuss how the PDCCH monitoring is
domain. On the other hand, for non-interleaved CCE-to-REG configured to a UE and how the UE deals with the cases when

CORESET (54 PRBs, 2 OFDM symbols)

Time

Frequency
Interleaved CCE-to-REG mapping using 27 x 2 block interleaver for 54 REG bundles

#0 #1 REG #17
REG bundle
REG = 2 REGs

CCE index #0 #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13 #14 #15 #16 #17

1 candidate for AL=4

2 candidates for AL=2

4 candidates for AL=1

Fig. 3 Example of PDCCH candidate determination and REG mapping.

 4

Frame 0 1 …
Slot 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 …

CSS set s = 0
24 24 24 24 24 24 24 24 …
(k=5, o=2, d=2)

USS set s = 2
20 24 28 18 18 27 40 26 30 34 31 30 31 19 27 32 23 41 32 26 …
(k=1, o=0, d=1)

USS set s = 3
16 18 15 17 24 20 21 19 14 28 14 23 24 16 17 22 15 19 20 23 …
(k=1, o=0, d=1)

Total number of CCEs that require channel estimation exceed the limit of 56 on the slot
è PDCCH candidates on the highlighted USS set are not mapped

Fig. 4 Number of CCEs that require channel estimations.

the UE is configured with more CCEs or blind decodes than the

B. PDCCH Candidate Hash Function
UE is capable of monitoring.
The mapping of PDCCH candidates of an SS set to CCEs of
A. Search Space Sets the associated CORESET is implemented by means of a hash
The UE performs blind decoding for a set of PDCCH function. The hash function randomizes the allocation of the
candidates. PDCCH candidates to be monitored are configured PDCCH candidates within CORESET p in slot 𝑛𝑠 . This is done
for a UE by means of search space (SS) sets. There are two SS according to

( 
set types: common SS (CSS) set, which is commonly monitored
by a group of UEs in the cell, and UE-specific SS (USS) set, )
L Yp ,ns + j (pL,m) mod  NCCE, p / L  + i,
which is monitored by an individual UE. A UE can be
configured with up to 10 SS sets each for up to four BWPs in a where a single carrier operation with a single SS set with index
serving cell. Therefore, a UE can be configured with up to 40 s is assumed for simplicity. The following are also assumed.
SS sets, where each has an index of 0-39. These numbers were • L is the aggregation level,
selected sufficiently large to cover all scenarios envisioned by • 𝑁CCE,p is the total number of CCEs for given CORESET p,
the industries so far. In general, SS set configuration provides a • m (0, 1, …, 𝑀(𝐿) − 1) is the candidate index with 𝑀(𝐿)
UE with the SS set type (CSS set or USS set), DCI format(s) to being the number of PDCCH candidates for AL L,
be monitored, monitoring occasion, and the number of PDCCH • i (0, 1, …, 𝐿 − 1) is the contiguous CCE index of the
candidates for each AL in the SS set. There is a special CSS set PDCCH candidate,
with index 0 (SS set 0), which is configured using a four-bit (𝐿) 𝑚∙𝑁𝐶𝐶𝐸,𝑝
• 𝑗𝑝,𝑚 = ⌊ (𝐿) ⌋,
information element in the MIB with respect to cell-defining 𝐿∙𝑀
SSB. The SS set 0 can be monitored even before higher-layer • 𝑌𝑝,𝑛𝑠 = 0 for the CSS set,
configurations are provided, while its configuration is restricted • 𝑌𝑝,𝑛𝑠 (0, 1, …, 216 - 1) for the USS set is a pseudo-random
to a limited number of combinations of parameters compared to variable based on the C-RNTI of the UE and slot number
SS sets of other than 0. 𝑛𝑠 , and
A SS set with index s is associated with only one CORESET • ⌊.⌋ denotes the floor operation.
with index p. This mapping from a SS set to a CORESET
enables simpler realization of multiple CORESETs and One example of an SS set determination in a CORESET,
multiple SS sets in a cell. The UE determines the slot for taking into account the CCE-to-REG mapping and the hash
monitoring the SS set with index s based on the higher layer function, is illustrated in Fig. 3. The CORESET has 54 PRBs
parameters for periodicity k, offset o, and duration d, where and two OFDM symbols with REG bundle size of two REGs.
periodicity k and offset o provide a starting slot and duration d The SS set is assumed to have M(L) = (4, 2, 1, 0, 0) for AL L =
provides the number of consecutive slots where the SS set is (1, 2, 4, 8, 16), respectively. For CCE-to-REG mapping, block
monitored starting from the slot identified by k and o. interleaver with the row size of two is used. The NR hash
Monitoring occasions of a SS set with index s within the slot function stems from LTE hash function design with higher
are configured by a bitmap parameter in the SS set configurability; other designs targeting further optimization,
configuration. Each bit representing an OFDM symbol within such as nested hash function from [5] or pseudo-random hash
the slot corresponds to the first OFDM symbol of the function from [9], was not adopted in NR.
monitoring occasion of the SS set.
 5

COR COR COR COR COR COR

ESE ESE ESE ESE ESE ESE
T#1 T#2 T#1 T#1 T#2 T#1

SS SS SS SS SS SS
set set set set set set
#1 #2 #3 #1 #2 #3

slot#0 slot#1 slot#2 slot#3

Fig. 5 Example of multi-beam operation for PDCCH

𝐷𝐿,𝜇
C. UE Monitoring Capabilities and Overbooking A BS configures a UE with 𝑁𝑐𝑒𝑙𝑙𝑠 serving cells, which does
not exceed the UE capability of CA configuration. The UE
Monitoring a large number of PDCCH candidates or CCEs
monitors the PDCCH for each serving cell, where the non-CA
increases the UE complexity. Therefore, NR specifies the
limit for each serving cell still holds. However, for a CA with a
maximum number of PDCCH candidates that require blind large number of serving cells, the UE implementation
𝑚𝑎𝑥,𝜇
decodes, 𝑀𝑃𝐷𝐶𝐶𝐻 = {44, 36, 22, 20}, and the maximum number complexity or power consumption for PDCCH monitoring may
𝑚𝑎𝑥,𝜇
of CCEs that require channel estimations, 𝐶𝑃𝐷𝐶𝐶𝐻 = {56, 56, 48, be problematic. Therefore, NR allows the UE to report the
32}, that UEs are capable of handling in a slot of SCS 𝜇 = limits for CA (CA-limit), which is less than the non-CA limit
{0, 1, 2, 3}. This limits the UE complexity to a reasonable level multiplied by the number of configured serving cells. In
with an acceptable restriction on the configuration of SS sets particular, a UE may report the number of serving cells that it
for PDCCH monitoring. 𝑐𝑎𝑝
is capable of monitoring 𝑁𝑐𝑒𝑙𝑙𝑠 ≥ 4. This means that when a BS
When a UE is configured with more than one SS set, the 𝐷𝐿,𝜇
configures the number of serving cells 𝑁𝑐𝑒𝑙𝑙𝑠 ≤ 4 or the UE is
number of PDCCH candidates/CCEs varies across slots due to
capable of monitoring more cells than the BS configures
independent monitoring occasions for the SS sets and slot- 𝐷𝐿,𝜇 𝑐𝑎𝑝
dependent hashing of different ALs for each SS set. Therefore, 𝑁𝑐𝑒𝑙𝑙𝑠 ≤ 𝑁𝑐𝑒𝑙𝑙𝑠 , the UE capability is a non-CA limit on each of
𝑐𝑎𝑝
a base station (BS) is allowed to configure the UE with a the serving cells. For the remaining case when 4 < 𝑁𝑐𝑒𝑙𝑙𝑠 <
𝐷𝐿,𝜇
number of PDCCH candidates/CCEs per slot that exceeds the 𝑁𝑐𝑒𝑙𝑙𝑠 , the CA-limit is applied. The BS takes into account the
UE capability, which is referred to as overbooking. Based on UE capability to configure SS sets for CA.
𝑐𝑎𝑝
the configuration, the UE and BS map PDCCH candidates in Note that the UE reports a single value for 𝑁𝑐𝑒𝑙𝑙𝑠 , even if the
each slot according to the following mapping rules: (i) CSS sets UE supports CA with various SCS combinations across serving
are mapped before USS sets, (ii) USS sets are mapped in cells where the UE monitors PDCCH candidates. For CA with
ascending order of the SS set indices, and if the number of different SCSs across serving cells, the serving cells are
PDCCH candidates/CCEs exceeds either of the UE processing grouped based on their SCS 𝜇, and the total number of PDCCH
limits, then (iii) No more SS sets are mapped in the slot after candidates that require blind decoding is split into the groups of
reaching the UE processing limit. These rules allow BS to cells of the same SCS. In particular, the maximum number of
number UE’s SS sets according to desired priority. PDCCH candidates that requires blind decoding, per group of
An example of mapping SS sets is illustrated in Fig. 4 cells of SCS 𝜇, i.e. the CA-limit, is defined as
assuming SCS 𝜇 = 0 . In this example, only the number of 𝑡𝑜𝑡𝑎𝑙,𝜇 𝑐𝑎𝑝 𝑚𝑎𝑥,𝜇 𝐷𝐿,𝜇 𝐷𝐿,𝑗
CCEs is illustrated for simplicity. We assume one CSS set 𝑠 = 𝑀𝑃𝐷𝐶𝐶𝐻 = ⌊𝑁𝑐𝑒𝑙𝑙𝑠 𝑀𝑃𝐷𝐶𝐶𝐻 𝑁𝑐𝑒𝑙𝑙𝑠 / ∑3𝑗=0 𝑁𝑐𝑒𝑙𝑙𝑠 ⌋.
0, is configured consuming 24 constant CCEs in CORESET Similarly, NR defines the CA-limit for the maximum number
𝑝 = 0. CSS hashing does not depend on slot number 𝑛𝑠 . In of CCEs that require channel estimations per group of cells of
CORESET 𝑝 = 1 , two USS sets, 𝑠 = 2 and 𝑠 = 3 , are SCS 𝜇, as
configured, and their CCE consumption levels vary due to slot-
𝑡𝑜𝑡𝑎𝑙,𝜇 𝑐𝑎𝑝 𝑚𝑎𝑥,𝜇 𝐷𝐿,𝜇 𝐷𝐿,𝑗
dependent hashing of different ALs. When the total number of 𝐶𝑃𝐷𝐶𝐶𝐻 = ⌊𝑁𝑐𝑒𝑙𝑙𝑠 𝐶𝑃𝐷𝐶𝐶𝐻 𝑁𝑐𝑒𝑙𝑙𝑠 / ∑3𝑗=0 𝑁𝑐𝑒𝑙𝑙𝑠 ⌋.
CCEs consumed by the CSS set and USS sets exceeds 56, the
Within a group of cells of the same numerology 𝜇, the limits,
USS set 𝑠 = 3 is not mapped. Without the overbooking feature, 𝑡𝑜𝑡𝑎𝑙,𝜇 𝑡𝑜𝑡𝑎𝑙,𝜇
the configuration for SS sets would be restricted by the worst- 𝑀𝑃𝐷𝐶𝐶𝐻 and 𝐶𝑃𝐷𝐶𝐶𝐻 , can be shared among cells, however, still
case number and hence a BS is unable to use the UE monitoring subject to non-CA limit per cell. Finally, the overbooking limit
capabilities to their maximum in all the slots. This results in is defined as the minimum between CA-limit and non-CA-limit
𝑡𝑜𝑡𝑎𝑙,𝜇 𝑚𝑎𝑥,𝜇 𝑡𝑜𝑡𝑎𝑙,𝜇 𝑚𝑎𝑥,𝜇
potential multi-user candidate blocking. as 𝑚𝑖𝑛(𝑀𝑃𝐷𝐶𝐶𝐻 , 𝑀𝑃𝐷𝐶𝐶𝐻 ) and 𝑚𝑖𝑛(𝐶𝑃𝐷𝐶𝐶𝐻 , 𝐶𝑃𝐷𝐶𝐶𝐻 ) . Note
that for CA, overbooking is allowed only on the primary cell of
D. UE Monitoring Capabilities and Overbooking for CA the cells group where in general various SS sets, e.g., CSS sets
𝑚𝑎𝑥,𝜇 𝑚𝑎𝑥,𝜇
Capabilities 𝑀𝑃𝐷𝐶𝐶𝐻 and 𝐶𝑃𝐷𝐶𝐶𝐻 defined in the previous and USS sets, are monitored. For a secondary cell where
subsection are referred to as a non-CA limit. In this subsection relatively fewer SS sets are monitored, the BS ensures that the
we introduce capabilities for CA, referred to as CA limit. maximum numbers of PDCCH candidates and non-overlapping
CCEs do not exceed the overbooking limits.
 6

IV. BEAMFORMING OPERATION TypeD corresponds to the receiver beam information on the UE
To overcome serious propagation loss and blockage for high side. If the DMRS of the PDCCH is quasi co-located with a RS
frequency ranges such as millimeter wave, NR can utilize with QCL-TypeD, this means that the UE may use the same
highly directional beamformed transmission and reception spatial reception parameters used for receiving the RS in analog
using a large number of antennas between the BS and UE [12]. beamforming to receive the PDCCH.
Hybrid beamforming is a combination of analog beamforming The BS can explicitly configure the QCL relationships using
that applies different phase shifters and/or amplitude weights higher layer signaling. The UE can be configured with up to 64
on each antenna panel and digital beamforming that applies TCI states for a CORESET to receive a PDCCH. Each TCI state
different digital precoders across panels. Since the analog contains parameters on RS resources and the QCL relationship
beamforming can only cover a limited area due to a narrow between the RS(s) and the DMRS ports of the PDCCH with
beam width, multiple beams need to be utilized to cover the respect to QCL-TypeA and QCL-TypeD. To receive PDCCH
entire cell. at a time, only one beam is used. Therefore, if more than one
To support multi-beam based operation for the PDCCH, NR TCI state is configured for a CORESET, the BS activates one
supports a higher-layer configuration for beamforming of the TCI states used for the CORESET using a medium access
(transmission configuration indication (TCI) state control (MAC) control element (CE) activation command [13].
configuration) per CORESET. When a UE monitors a SS set If a TCI state has not been indicated to a UE, the UE monitors
associated with a CORESET, the UE can receive the PDCCH the PDCCH in monitoring occasions of SS sets using the
in the CORESET based on the TCI state configuration receiver beam that the UE used for receiving the SSB selected
configured for the CORESET as illustrated in Fig. 5. during the initial access procedure. In other words, the UE can
In the following subsections, we discuss more details on the assume that the DMRS antenna port associated with the
PDCCH beamforming operation. PDCCH in the CORESET is quasi co-located with the
corresponding SSB with respect to QCL-TypeA and QCL-
A. Beam Configuration Aspects TypeD properties.
Beam information for PDCCH reception is implicitly known
B. Beam Failure Recovery Aspects
to the UE by means of the quasi-co-location (QCL) relationship
between DL reference signals (RSs) (typically channel state NR supports link recovery procedure based on layer 1
information RS (CSI-RS), each of which is associated with each reference signal received power (L1-RSRP) measurement,
beam) and the DMRS of the PDCCH. In NR, the DMRS of the which is referred to as the beam failure recovery (BFR)
PDCCH can be quasi co-located with a RS with QCL-TypeA procedure. That is, once there is a misunderstanding regarding
and/or QCL-TypeD. QCL-TypeA corresponds to the channel the appropriate beams between the BS and UE that leads to poor
statistical properties observed on the UE side including Doppler link quality, the UE can indicate a preferred beam to the BS by
shift, Doppler spread, average delay, and delay spread. QCL- transmitting a physical random access channel (PRACH) on the
resource associated with the preferred beam.

Table 1 Summary of DCI formats in 5G NR

DCI format Usage RNTI for CRC scrambling

DCI 0_0 Fallback DCI format for scheduling of PUSCH • C-RNTI or MCS-C-RNTI for dynamically scheduled unicast
transmission
• CS-RNTI for configured scheduled unicast transmission
• TC-RNTI for Message 3 transmission in contention based random access
DCI 0_1 Non-fallback DCI format for scheduling of • C-RNTI or MCS-C-RNTI for dynamically scheduled unicast
PUSCH transmission
• CS-RNTI for configured scheduled unicast transmission
• SP-CSI-RNTI for activation of semi-persistent channel state information
reporting on PUSCH
DCI 1_0 Fallback DCI format for scheduling of PDSCH • C-RNTI or MCS-C-RNTI for dynamically scheduled unicast
transmission
• CS-RNTI for configured scheduled unicast transmission
• SI-RNTI for broadcasting of system information
• P-RNTI for paging and system information change notification
• RA-RNTI for random access response
• TC-RNTI for contention resolution (when no valid C-RNTI is available)
DCI 1_1 Non-fallback DCI format for scheduling of • C-RNTI or MCS-C-RNTI for dynamically scheduled unicast
PDSCH transmission
• CS-RNTI for configured scheduled unicast transmission
DCI 2_0 Notifying a group of UEs of the slot format • SFI-RNTI for slot format indication
DCI 2_1 Notifying a group of UEs of the PRB(s) and • INT-RNTI for pre-emption indication in downlink
OFDM symbol(s) where UE may assume no
transmission is intended for the UE
DCI 2_2 Transmission of a group of power control • TPC-PUCCH-RNTI for PUCCH power control
commands for PUCCH and PUSCH • TPC-PUSCH-RNTI for PUSCH power control
DCI 2_3 Transmission of a group of SRS requests and • TPC-SRS-RNTI for SRS trigger and power control
power control commands for SRS transmissions
 7

A UE can be configured with a set of CSI-RS resources (a set information change notification, random access response, and
q0) for the purpose of beam failure detection. Then the UE contention resolution (when no valid C-RNTI is available).
measures the link quality according to the CSI-RS resources in Uplink scheduling grants are supported by DCI format 0_0
set q0. If the measured radio link quality for all corresponding (known as the fallback format) and DCI format 0_1 (known as
CSI-RS resources in set q0 is worse than the pre-defined the non-fallback format). Similar to the DCI formats used for
threshold, the UE can request BFR by transmitting the PRACH DL scheduling assignments, DCI format 0_0 supports a basic
corresponding to the selected qnew from the candidate SSB or set of NR functionality while DCI format 1_1 supports all NR
CSI-RS set q1 which is configured for BFR. After transmitting features.
the PRACH, the UE monitors PDCCH in a search space In addition to the DCI formats used for DL scheduling
assignments and UL scheduling grants, there are DCI formats
configured for BFR (SS-BFR) to receive the corresponding
2_0, 2_1, 2_2, and 2_3 introduced for special purposes of
response. Once a PDCCH with the response is successfully
signaling to a configured group of one or more devices. The
received from the SS-BFR, the UE considers that the BFR is
CRCs of these DCI formats are scrambled by specific RNTIs,
successful. as described in Table 1. DCI format 2_0 is used for slot format
During the PDCCH monitoring in the SS-BFR, the UE indication that dynamically determines whether the resources
assumes that the PDCCH DMRS is quasi co-located with the are used for UL or DL transmission. DCI format 2_1 is used for
selected CSI-RS or SS/PBCH block with index qnew. The UE preemption indication that signals that some resources have
can receive the physical downlink shared channel (PDSCH) been preempted and thus are not used for transmission to the
(scheduled by the PDCCH transmitted through the SS-BFR) notified device(s). DCI format 2_2 is used for transmitting
providing the MAC CE activation command for TCI state physical uplink shared channel (PUSCH) or physical uplink
update/change for the CORESETs. control channel (PUCCH) power-control commands. The main
use case is to support power control for semi-persistent or
C. Beam Collision Aspects
periodic uplink transmissions. DCI format 2_3 is used for
The number of analog beams that can be received transmitting sounding reference signal (SRS) power-control
simultaneously is restricted by the number of antenna panels commands and optionally SRS requests.
supported on the UE side since analog beamforming is applied Different DCI formats may or may not have the same DCI
per antenna panel. In NR Release 15, the study focused on a size. An NR UE is capable of monitoring up to three different
single panel scenario, so that a UE can receive the physical DCI sizes using the C-RNTI. Additionally, the UE is capable of
channels at the same time using a single analog beam. Taking monitoring one additional DCI size using RNTIs for special
into account this restriction, some PDCCH monitoring rules are purposes (such as SFI-RNTI and INT-RNTI). The DCI
defined in NR. scrambled with the C-RNTI (MCS-C-RNTI and CS-RNTI) is
When some PDCCH monitoring occasions in multiple time critical and requires the UE to decode it promptly to
CORESETs configured with different QCL-TypeD overlap in process the scheduled data transmission. The DCI scrambled
the time domain, the UE monitors PDCCHs only in the with a RNTI used for a special purpose is less time-critical for
CORESET configured to monitor the CSS set with the lowest the UE to decode. This requirement is usually referred to as a
index. If all the overlapped PDCCH monitoring occasions are “3+1” DCI size budget.
not associated with any CSS set, the PDCCHs in the CORESET Due to the constraint of the DCI size budget, the sizes of
some DCI formats need to be aligned by padding, truncation,
configured with the USS set with the lowest index are
and/or determining the frequency domain resource assignment
monitored.
field differently.
V. DOWNLINK CONTROL INFORMATION VI. PDCCH ENHANCEMENTS IN RELEASE 16
In this section, we discuss the details of DCI which is the The first 5G NR specifications were completed in Release
payload transmitted on a PDCCH. 15. 3GPP has started to evolve NR in Release 16 [14] to further
A UE identifies the payload size and RNTI used for CRC improve performance and address new use cases. PDCCH
scrambling for a DCI format as presented in Section II. There enhancements are an important part of Release 16 and span
are various DCI formats and RNTIs, as summarized in Table 1. across multiple topics, as overviewed below. In Release 16,
Downlink scheduling assignments are supported by DCI physical layer enhancements for NR URLLC is being specified.
format 1_0 (known as the fallback format) and DCI format 1_1 For some URLLC services, frequent PDCCH monitoring
(known as the non-fallback format). The design rationale is that and/or reliability enhancements to PDCCH are required to meet
the fallback format as a default scheduling option should be the latency and reliability requirements. 3GPP Release 16 is
robust to support the basic NR operations while the non- considering to enhance the PDCCH in terms of both latency and
fallback format should be flexible to accommodate the rich set reliability viewpoints. As for the latency enhancements,
of NR features. Specifically, DCI format 1_0 supports a basic PDCCH monitoring capability can be re-defined and upgraded;
set of NR functionality and its fields are generally not it can change from “per slot” to “per monitoring span” and
configurable. DCI format 1_1 supports all NR features and its specify higher values for Release 16 compared to Release 15.
fields are highly configurable. In general, DCI format 1_0 is As for the reliability enhancements, various DCI fields for
smaller than DCI format 1_1. DCI format 1_0 is particularly PDCCH would become variable fields. With this, DCI message
needed for scheduling system information, paging and system payload can be controlled by network such that it can be very
 8

compact and hence, the system can take a good balance between dynamically control the UE’s PDCCH monitoring behavior
higher performance and higher flexibility. depending on the data traffic. The power saving signal is
In vehicular-to-everything (V2X), the Physical Sidelink transferred by PDCCH of which the monitoring occasions are
Control Channel (PSCCH) is defined as the counterpart of located before discontinuous reception (DRX) cycle. The
PDCCH for control information transmission between UEs. power saving signal can indicate whether the UE skips the
Correspondingly, the Sidelink Control Information (SCI) is subsequent monitoring occasions within the DRX on duration
defined as the counterpart of DCI for PDCCH. Unlike PDCCH, or not. By this way, the UE will wake up to monitor PDCCH
PSCCH supports a 2-stage control channel design. Decoding of only when there is scheduled data. In [15], it was evaluated that
the first stage SCI requires blind detection. The first stage SCI a remarkable power saving gain can be achieved with endurable
only needs to carry the basic information such as resource latency loss under the low data traffic circumstance.
allocation for the data channel and the second stage SCI. This
design seems less flexible than PDCCH but it provides enough VII. CONCLUSIONS
flexibility for the broadcast operation of the sidelink. The This article presented an overview of the 5G NR PDCCH
second stage SCI carries additional information that is not used
with details on the mechanisms including physical layer
for scheduling and it does not need blind detection. With this
structure, monitoring schemes, beamforming operation, and the
design, the number of blind detections for the first stage PSCCH
carried information. The design of the 5G NR PDCCH enables
decoding can be less than the PDCCH for which separate blind
detections are needed for DCI formats with different sizes. As various use-cases and operations in a wide range of carrier
a result, the processing complexity of PSCCH decoding can be frequencies.
reduced.
In NR unlicensed (NR-U) in sub-7GHz, a BS monitors the REFERENCES
wireless channel and checks whether it has gained the right to [1] http://www.3gpp.org/news-events/3gpp-news/1965-rel-15_news, Jan.
use the channel based on Listen Before Talk (LBT) mechanism. 2019.
[2] 3GPP TS22.261, “Service requirements for 5G system; Stage 1,” v15.0.0,
LBT is performed in granularity of 20 MHz sub-bands. Once Mar. 2017.
LBT passes for a sub-band, the BS can transmit (PDCCH, data, [3] 3GPP TR38.913, “Study on scenarios and requirements for next
etc.) on the sub-band to UE(s). When UE’s BWP comprises generation access technologies,” v15.0.0, July 2018.
multiple 20 MHz sub-bands, specifications allow to confine a [4] X. Lin, et al., “5G new radio: Unveiling the essentials of the next
generation wireless access technology,” arXiv preprint arXiv:1806.06898,
PDCCH within an LBT sub-band to avoid partial puncturing of to appear in IEEE Communications Standards Magazine, June 2018.
PDCCH candidates when LBT fails for some sub-bands of the [5] 3GPP TS38.211, “NR; Physical channels and modulation,” v15.3.0, Sept.
BWP. Confinement is enabled by a configuration of a NR-U 2018.
CORESET within a sub-band that can be replicated in the sub- [6] 3GPP TS38.213, “NR; Physical layer procedure for control,” v15.3.0,
Oct.2018.
bands of the BWP, as if the same CORESET would be [7] 3GPP TS38.214, “NR; Physical layer procedure for data,” v15.3.0,
separately configured in those sub-bands. This reduces the Oct.2018.
complexity of handling additional CORESETs, as these [8] Fatemeh Hamidi-Sepehr; et al., “5G NR PDCCH: Design and
CORESETs share the same set of parameters and the same QCL Performance”, 2018 IEEE 5G World Forum (5GWF), July 2018
[9] Volker Braun, et al., “5G NR Physical Downlink Control Channel:
assumption. Additionally, the configuration overhead is Design, Performance and Enhancements”, IEEE Wireless
reduced. NR-U will likely use the NR Release 15 DCI formats Communications and Networking Conference, Apr. 2019
enhanced with NR-U specific fields to transmit control [10] Erik Dahlman et al. " 5G NR: The Next Generation Wireless Access
information. For example, an additional field is added to NR Technology", Academic Press, Aug 2018
[11] 3GPP TS38.212, “NR; Channel coding and multiplexing,” v15.3.0, Sept.
DCI format with CRC scrambled by SFI-RNTI to indicate sub- 2018.
bands of a BWP on which the BS is transmitting. [12] E. Onggosanusi et al, “Modular and high-resolution channel state
Enhanced MIMO in Release 16 includes support of multiple information and beam management for 5G new radio,” IEEE
transmit receive point (multi-TRP) transmission. In multi-TRP, Communications Magazine, vol. 56, no 3, pp. 48 -55, March 2018.
[13] 3GPP TS38.321, “NR; Medium Access Control (MAC) protocol
different data from multiple TRPs may be transmitted for specification” v15.3.0, Sept. 2018.
spatial multiplexing to increase data rate, or same data from [14] 3GPP TR 21.916, “Summary of Rel-16 Work Items,” v0.1.0, Sept. 2019.
multiple TRPs may be transmitted for diversity to improve [15] 3GPP TR 38.840, “Study on UE Power Saving,” v2.0.0, May 2019.
transmission reliability and robustness. To efficiently support
multi-TRP, both single PDCCH and multiple PDCCH based
schemes are being introduced in Release 16. In single PDCCH
based scheme, one TRP sends PDCCH to schedule one set of
PDSCH layers from a first TRP and a second set of PDSCH
layers from a second TRP. In multiple PDCCH based scheme,
two TRPs independently schedule PDSCHs from two TRPs.
From UE power saving perspective, power saving techniques
to reduce the UE’s PDCCH monitoring operations is under
discussion in Release 16. It was shown that a large amount of
UE’s power in modem composed of radio frequency (RF) chain
and base band (BB) unit is consumed due to the PDCCH
monitoring without any scheduled data [15]. Motivated by this,
a power saving signal (a.k.a wake-up signal) is introduced to