Skip navigation

Designing for the future: the 5G NR physical layer

Future networks will have to provide broadband access wherever needed and support a diverse range of services including everything from robotic surgery to virtual reality classrooms and self-driving cars. 5G New Radio is designed to fit these requirements, with physical layer components that are flexible, ultra-lean and forward-compatible.

Ericsson Technology Review logo


Authors: Ali A. Zaidi, Robert Baldemair, Mattias Andersson, Sebastian Faxér, Vicent Molés-Cases, Zhao Wang

Download pdf 

BPSK - binary phase shift keying
BS - base station
CDM - code division multiplexing
CPE - common phase error
CP-OFDM - cyclic prefix orthogonal frequency division multiplexing
CSI-RS - channel-state information reference signal
D2D - device-to-device
DFT-SOFDM - discrete Fourier transform spread orthogonal frequency division multiplexing
DL - downlink
DMRS - demodulation reference signal
eMBB - enhanced mobile broadband
eMBMS - evolved multimedia broadcast multicast service
FDM - frequency division multiplexing
HARQ - hybrid automatic repeat request
IMT-2020 - the next generation mobile communication systems to be specified by ITU-R
IoT - Internet of Things
ITU-R - International Telecommunication Union Radiocommunication Sector
LBT - listen-before-talk
LDPC - low-density parity-check
MBB - mobile broadband
MIMO - multiple-input, multiple-output
mMTC - massive machine-type communications
mmWave - millimeter wave
MU-MIMO - multi-user MIMO
NR - New Radio
PRB - physical resource block
PTRS - phase-tracking reference signal
QAM - quadrature amplitude modulation
QPSK - quadrature phase shift keying
SRS - sounding reference signal
TDM - time division multiplexing
UE - user equipment
UL - uplink
URLLC - ultra-reliable low-latency communications

Far more than an evolution of mobile broad­band, 5G wireless access will be a key IoT enabler, empowering people and industries to achieve new heights in terms of efficiency and innovation. A recent survey performed across eight different industries (automotive, finance, utilities, public safety, health care, media, internet and manufacturing) revealed that 89 percent of respondents expect 5G to be a game changer in their industry [1].

5G wireless access is being developed with three broad use case families in mind: enhanced mobile broadband (eMBB), massive machine-type communications (mMTC) and ultra-reliable low-latency communications (URLLC) [2, 3]. eMBB focuses on across-the-board enhancements to the data rate, latency, user density, capacity and coverage of mobile broadband access. mMTC is designed to enable communication between devices that are low-cost, massive in number and battery-driven, intended to support applications such as smart metering, logistics, and field and body sensors. Finally, URLLC will make it possible for devices and machines to communicate with ultra-reliability, very low latency and high availability, making it ideal for vehicular communication, industrial control, factory automation, remote surgery, smart grids and public safety applications. 

To meet the complex and sometimes contradictory requirements of these diverse use cases, 5G will encompass both an evolution of today’s 4G (LTE) networks and the addition of a new, globally standardized radio access technology known as New Radio (NR).

Change to 5G New Radio

5G NR will operate in the frequency range from below 1GHz to 100GHz with different deployments. There will typically be more coverage per base station (macro sites) at lower carrier frequencies, and a limited coverage area per base station (micro and pico sites) at higher carrier frequencies. To provide high service quality and optimal reliability, licensed spectrum will continue to be the backbone of the wireless network in 5G, and transmission in unlicensed spectrum will be used as a complement to provide even higher data rates and boost capacity. The overall vision for 5G in terms of use cases, operating frequencies and deployments is shown in Figure 1.

Figure 1

Figure 1: 5G vision: use cases, spectrum and deployments

The standardization of NR started in 3GPP in April 2016, with the aim of making it commercially available before 2020. 3GPP is taking a phased approach to defining the 5G specifications. A first standardization phase with limited NR functionality will be completed by 2018, followed by a second standardization phase that fulfills all the requirements of IMT-2020 (the next generation of mobile communication systems to be specified by ITU-R) by 2019. It is likely that NR will continue to evolve beyond 2020, with a sequence of releases including additional features and functionalities. Although NR does not have to be backward compatible with LTE, the future evolution of NR should be backward compatible with its initial release(s). Since NR must support a wide range of use cases – many of which are not yet defined – forward compatibility is of utmost importance.

NR physical layer design

A physical layer forms the backbone of any wireless technology. The NR physical layer has a flexible and scalable design to support diverse use cases with extreme (and sometimes contradictory) requirements, as well as a wide range of frequencies and deployment options.

The key technology components of the NR physical layer are modulation schemes, waveform, frame structure, reference signals, multi-antenna transmission and channel coding.

Modulation schemes

LTE supports the QPSK, 16QAM, 64QAM and 256QAM modulation formats, and all of these will also be supported by NR. In addition, 3GPP has included л/2-BPSK in UL to enable a further reduced peak-to-average power ratio and enhanced power-amplifier efficiency at lower data rates, which is important for mMTC services, for example. Since NR will cover a wide range of use cases, it is likely that the set of supported modulation schemes may expand. For example, 1024QAM may become part of the NR specification, since fixed point-to-point backhaul already uses modulation orders higher than 256QAM. Different modulation schemes for different UE categories may also be included in the NR specification. 


3GPP has agreed to adopt CP-OFDM with a scalable numerology (subcarrier spacing, cyclic prefix) in both UL and DL up to at least 52.6GHz. Having the same waveform in both directions simplifies the overall design, especially with respect to wireless backhauling and device-to-device (D2D) communications. Additionally, there is support for DFT-Spread OFDM in UL for coverage-limited scenarios, with single stream transmissions (that is, without spatial multiplexing). Any operation that is transparent to a receiver can be applied on top of CP-OFDM at the transmitter side, such as windowing/filtering to improve spectrum confinement.  

A scalable OFDM numerology is required to enable diverse services on a wide range of frequencies and deployments. The subcarrier spacing is scalable according to 15×2n kHz, where n is an integer and 15kHz is the subcarrier spacing used in LTE. The scaling factor 2n ensures that slots and symbols of different numerologies are aligned in the time domain, which is important to efficiently enable TDD networks [4]. The details related to NR OFDM numerologies are shown in Figure 2. The choice of parameter n depends on various factors including type of deployment, carrier frequency, service requirements (latency, reliability and throughput), hardware impairments (oscillator phase noise), mobility and implementation complexity [5]. For example, wider subcarrier spacing can be promising for latency-critical services (URLLC), small coverage areas and higher carrier frequencies. Narrower subcarrier spacing can be utilized for lower carrier frequencies, large coverage areas, narrowband devices and evolved multimedia broadcast multicast services (eMBMSs). It may also be possible to support multiple services simultaneously with different requirements on the same carrier by multiplexing two different numerologies (wider subcarrier spacing for URLLC and lower subcarrier spacing for MBB/mMTC/eMBMS, for example). 

Figure 2

Figure 2: Scalable OFDM numerology for NR

The spectrum of OFDM signal decays rather slowly outside the transmission bandwidth. In order to limit out-of-band emission, the spectrum utilization for LTE is 90 percent. That is, 100 of the 111 possible physical resource blocks  (PRBs) are utilized in a 20MHz bandwidth allocation. For NR, it has been agreed that the spectrum utilization will be greater than 90 percent. Windowing and filtering operations are viable ways to confine the OFDM signal in the frequency domain. It is important to note that the relationship between spectrum efficiency and spectrum confinement is not linear, since spectrum confinement techniques can induce self-interference. 

Frame structure

NR frame structure supports TDD and FDD transmissions and operation in both licensed and unlicensed spectrum. It enables very low latency, fast HARQ acknowledgements, dynamic TDD, coexistence with LTE and transmissions of variable length (for example, short duration for URLLC and long duration for eMBB). The frame structure follows three key design principles to enhance forward compatibility and reduce interactions between different features.

The first principle is that transmissions are self-contained. Data in a slot and in a beam is decodable on its own without dependency on other slots and beams. This implies that reference signals required for demodulation of data are included in a given slot and a given beam.  

The second principle is that transmissions are well confined in time and frequency. Keeping transmissions together makes it easier to introduce new types of transmissions in parallel with legacy transmissions in the future. NR frame structure avoids the mapping of control channels across full system bandwidth. 

The third principle is to avoid static and/or strict timing relations across slots and across different transmission directions. For example, asynchronous HARQ is used instead of predefined retransmission time. 

As shown in Figure 2, a slot in NR comprises seven or 14 OFDM symbols for ≤ 60kHz numerologies and 14 OFDM symbols for ≥ 120kHz numerologies. A slot duration also scales with the chosen numerology since the OFDM symbol duration is inversely proportional to its subcarrier spacing. Figure 3 provides examples for TDD, with guard periods for UL/DL switching. 

Figure 3

Figure 3: TDD-based frame structure examples for eMBB, URLLC and operation in unlicensed spectrum using listen-before-talk (LBT)

A slot can be complemented by mini-slots to support transmissions with a flexible start position and a duration shorter than a regular slot duration. A mini-slot can be as short as one OFDM symbol and can start at any time. Mini-slots can be useful in various scenarios, including low-latency transmissions, transmissions in unlicensed spectrum and transmissions in the millimeter wave spectrum (mmWave band).  

In low-latency scenarios, transmission needs to begin immediately without waiting for the start of a slot boundary (URLLC, for example). When transmitting in unlicensed spectrum, it is beneficial to start transmission immediately after LBT. When transmitting in mmWave band, the large amount of bandwidth available implies that the payload supported by a few OFDM symbols is large enough for many of the packets. Figure 3 provides examples of URLLC- and LBT-based transmission in unlicensed spectrum via mini-slots and illustrates that multiple slots can be aggregated for services that do not require extremely low latency (eMBB, for example). Having a longer transmission duration helps to increase coverage or reduce the overhead due to switching (in TDD), transmission of reference signals and control information.

The same frame structure can be used for FDD, by enabling simultaneous reception and transmission (that is, DL and UL can overlap in time). This frame structure is also applicable to D2D communications. In that case, the DL slot structure can be used by the device that is initiating (or scheduling) the transmission, and the UL slot structure can be used by the device responding to the transmission. 

NR frame structure also allows for rapid HARQ acknowledgement, in which decoding is performed during the reception of DL data and the HARQ acknowledgement is prepared by the UE during the guard period, when switching from DL reception to UL transmission. 

To obtain low latency, a slot (or a set of slots in case of slot aggregation) is front-loaded with control signals and reference signals at the beginning of the slot (or set of slots). 

Reference signals

NR has an ultra-lean design that minimizes always-on transmissions to enhance network energy efficiency and ensure forward compatibility. In contrast to the setup in LTE, the reference signals in NR are transmitted only when necessary. The four main reference signals are the demodulation reference signal (DMRS), phase-tracking reference signal (PTRS), sounding reference signal (SRS) and channel-state information reference signal (CSI-RS).  

DMRS is used to estimate the radio channel for demodulation. DMRS is UE-specific, can be beamformed, confined in a scheduled resource, and transmitted only when necessary, both in DL and UL. To support multiple-layer MIMO transmission, multiple orthogonal DMRS ports can be scheduled, one for each layer. Orthogonality is achieved by FDM (comb structure) and TDM and CDM (with cyclic shift of the base sequence or orthogonal cover codes). The basic DMRS pattern is front loaded, as the DMRS design takes into account the early decoding requirement to support low-latency applications. For low-speed scenarios, DMRS uses low density in the time domain. However, for high-speed scenarios, the time density of DMRS is increased to track fast changes in the radio channel.

PTRS is introduced in NR to enable compensation of oscillator phase noise. Typically, phase noise increases as a function of oscillator carrier frequency. PTRS can therefore be utilized at high carrier frequencies (such as mmWave) to mitigate phase noise. One of the main degradations caused by phase noise in an OFDM signal is an identical phase rotation of all the subcarriers, known as common phase error (CPE). PTRS is designed so that it has low density in the frequency domain and high density in the time domain, since the phase rotation produced by CPE is identical for all subcarriers within an OFDM symbol, but there is low correlation of phase noise across OFDM symbols. PTRS is UE-specific, confined in a scheduled resource and can be beamformed. The number of PTRS ports can be lower than the total number of ports, and orthogonality between PTRS ports is achieved by means of FDM. PTRS is configurable depending on the quality of the oscillators, carrier frequency, OFDM subcarrier spacing, and modulation and coding schemes used for transmission. 

The SRS is transmitted in UL to perform CSI measurements mainly for scheduling and link adaptation. For NR, it is expected that the SRS will also be utilized for reciprocity-based precoder design for massive MIMO and UL beam management. It is likely that the SRS will have a modular and flexible design to support different procedures and UE capabilities. The approach for CSI-RS is similar. 

Multi-antenna transmissions

NR will employ different antenna solutions and techniques depending on which part of the spectrum is used for its operation. For lower frequencies, a low to moderate number of active antennas (up to around 32 transmitter chains) is assumed and FDD operation is common. In this case, the acquisition of CSI requires transmission of CSI-RS in the DL and CSI reporting in the UL. The limited bandwidths available in this frequency region require high spectral efficiency enabled by multi-user MIMO (MU-MIMO) and higher order spatial multiplexing, which is achieved via higher resolution CSI reporting compared with LTE.  

For higher frequencies, a larger number of antennas can be employed in a given aperture, which increases the capability for beamforming and MU-MIMO. Here, the spectrum allocations are of TDD type and reciprocity-based operation is assumed. In this case, high-resolution CSI in the form of explicit channel estimations is acquired by UL channel sounding. Such high-resolution CSI enables sophisticated precoding algorithms to be employed at the BS. This makes it possible to increase multi-user interference suppression, for example, but might require additional UE feedback of inter-cell interference or calibration information if perfect reciprocity cannot be assumed. 

For even higher frequencies (in the mmWave range) an analog beamforming implementation is typically required currently, which limits the transmission to a single beam direction per time unit and radio chain. Since an isotropic antenna element is very small in this frequency region owing to the short carrier wavelength, a great number of antenna elements is required to maintain coverage. Beamforming needs to be applied at both the transmitter and receiver ends to combat the increased path loss, even for control channel transmission. A new type of beam management process for CSI acquisition is required, in which the BS needs to sweep radio transmitter beam candidates sequentially in time, and the UE needs to maintain a proper radio receiver beam to enable reception of the selected transmitter beam.  

To support these diverse use cases, NR features a highly flexible but unified CSI framework, in which there is reduced coupling between CSI measurement, CSI reporting and the actual DL transmission in NR compared with LTE. The CSI framework can be seen as a toolbox, where different CSI reporting settings and CSI-RS resource settings for channel and interference measurements can be mixed and matched so they correspond to the antenna deployment and transmission scheme in use, and where CSI reports on different beams can be dynamically triggered. The framework also supports more advanced schemes such as multi-point transmission and coordination. The control and data transmissions, in turn, follow the self-contained principle, where all information required to decode the transmission (such as accompanying DMRS) is contained within the transmission itself. As a result, the network can seamlessly change the transmission point or beam as the UE moves in the network.

Channel coding

NR employs low-density parity-check (LDPC) codes for the data channel and polar codes for the control channel. LDPC codes are defined by their parity-check matrices, with each column representing a coded bit, and each row representing a parity-check equation. LDPC codes are decoded by exchanging messages between variables and parity checks in an iterative manner. The LDPC codes proposed for NR use a quasi-cyclic structure, where the parity-check matrix is defined by a smaller base matrix. Each entry of the base matrix represents either a ZxZ zero matrix or a shifted ZxZ identity matrix. 

Unlike the LDPC codes implemented in other wireless technologies, the LDPC codes considered for NR use a rate-compatible structure, as shown in  Figure 4. The light blue part (top left) of the base matrix defines a high rate code, at a rate of either 2/3 or 8/9. Additional parity bits can be generated by extending the base matrix and including the rows and columns marked in dark blue (bottom left). This allows for transmission at lower code rates, or for generation of additional parity bits such as those used for HARQ operation using incremental redundancy similar to LTE. Since the parity-check matrix for higher code rates is smaller, decoding latency and complexity decreases for high code rates. Along with the high degree of parallelism achievable through the quasi-cyclic structure, this allows for very high peak throughputs and low latencies. Further, the parity-check matrix can be extended to lower rates than the LTE turbo codes, which rely on repetition for code rates below 1/3. This allows the LDPC codes to achieve higher coding gains also at low coding rates, making them suitable for use cases requiring high reliability.

Figure 4

Figure 4: Structure of NR LDPC matrices

Polar codes will be used for layer 1 and layer 2 control signaling, except for very short messages. Polar codes are a relatively recent invention, introduced by Arıkan in 2008 [6]. They are the first class of codes shown to achieve the Shannon capacity with reasonable decoding complexity for a wide variety of channels. 

By concatenating the polar code with an outer code and keeping track of the most likely values of previously decoded bits at the decoder (the list), good performance is achieved at shorter block lengths, like those typically used for layer 1 and layer 2 control signaling. By using a larger list size, error correction performance improves, at the cost of higher complexity at the decoder. 


Flexibility, ultra-lean design and forward compatibility are the pillars on which all the 5G NR physical layer technology components (modulation schemes, waveform, frame structure, reference signals, multi-antenna transmission and channel coding)  are being designed and built. The high level of flexibility and scalability in 5G NR will enable it to meet the requirements of diverse use cases, including a wide range of carrier frequencies and deployment options. Its built-in forward compatibility will ensure that 5G NR can easily evolve to support any unforeseen requirements.

The authors

Ali A. Zaidi

Ali A. Zaidi

joined Ericsson in 2014, where he is currently a senior researcher working with concept development and standardization of radio access technologies (NR and LTE-Advanced Pro). His research focuses on mmWave communications, indoor positioning, device-to-device communications, and systems for intelligent transportation and networked control. He holds an M.Sc. and a Ph.D. in telecommunications from KTH Royal Institute of Technology in Stockholm, Sweden. Zaidi is currently serving as a member of the Technology Intelligence Group Radio and a member of the Young Advisory Board at Ericsson Research.  

Robert Baldemair

Robert Baldemair

is a master researcher who joined Ericsson in 2000. He spent several years working with research and development of radio access technologies for LTE before shifting his focus to wireless access for 5G. He holds a Dipl. Ing. and a Ph.D. from the Technische Universität Wien in Vienna, Austria. He received the Ericsson Inventor of the Year award in 2010, and in 2014 he and a group of colleagues were nominated for the European Inventor Award.

Mattias Andersson

Mattias Andersson

is an experienced researcher who joined Ericsson in 2014. His work focuses on concept development and standardization of low-latency communications, carrier aggregation and channel coding for LTE and NR. He holds both an M.Sc. in engineering physics and a Ph.D. in telecommunications from KTH Royal Institute of Technology in Stockholm, Sweden.

Sebastian Faxér

Sebastian Faxér

is a researcher at Ericsson Research. He received an M.Sc. in applied physics and electrical engineering from Linköping University, Sweden, in 2014 and joined Ericsson the same year. Since then he has worked on concept development and standardization of multi-antenna technologies for LTE and NR.

Vicent Molés-Cases

Vicent Molés-Cases

is a researcher at Ericsson Research. He received an M.Sc. in telecommunication engineering from the Polytechnic University of Valencia in Spain in 2016 and joined Ericsson the same year. Since then he has worked on concept development and 3GPP standardization of reference signals for NR.

Zhao Wang

Zhao Wang

is a researcher at Ericsson Research who joined the company in 2015. He holds a Ph.D. in telecommunications from KTH Royal Institute of Technology in Stockholm, Sweden. His work currently focuses on reference signal design of NR for 3GPP standardizations.