Introduction
New waveforms must support multiple services on a single contiguous spectrum block (eMBB, mMTC, and URLLC) while accommodating a range of user velocities. Different service requirements, user speeds, and deployment scenarios can lead to different optimal algorithms. Therefore, 5G waveform design should support different numerologies, including different CP lengths, subcarrier spacings, and TTIs, and enable effective coexistence of different numerologies within a single contiguous spectrum block. For low-cost devices, the preferred waveform should also support asynchronous uplink access with some relaxation of synchronization.
To reflect these requirements, waveform evaluation should include representative 5G algorithms, coexistence scenarios, and asynchronous uplink transmission scenarios. Thus, link-level simulations should consider both downlink single-numerology and mixed-numerology cases, and uplink should consider asynchronous transmission performance.
Evaluation Scope and Cases
Evaluation requires a set of representative numerology parameters. Short CP and short TTI lengths should be considered in particular, since many new waveforms are sensitive to these parameters.
Downlink: Single-numerology case
The single-numerology case represents a carrier bandwidth configured with only one numerology. Its purpose is to evaluate the waveform's inherent intersymbol interference (ISI) and inter-carrier interference (ICI), and the system guard-band overhead required for various numerologies.
Downlink: Mixed-numerology case
This case represents a target user in one subband while adjacent subbands use different numerologies. Typically assume three subbands: the center subband is the target user, with two adjacent subbands using different numerologies. The mixed-numerology case evaluates the waveform's robustness to adjacent subband interference and the guard-band overhead required for effective coexistence on a single contiguous spectrum block.
Uplink: Asynchronous transmission case
This case represents a target UE with two adjacent interfering UEs having different power offsets. Typically assume three UEs: the center UE is the target, with two interfering UEs on either side at different power offsets. The objective is to evaluate waveform robustness to uplink asynchronous interference and to determine the guard-band overhead between UEs given timing advance relaxation or free uplink access for some devices.
For downlink, waveform design should be evaluated together with technologies that improve spectral efficiency, such as MIMO and higher-order modulation (e.g., 64QAM and 256QAM). For uplink asynchronous transmission, a mid-range MCS (e.g., 16QAM) and SISO are reasonable evaluation configurations.
Evaluation Metric: User Spectral Efficiency
User spectral efficiency is recommended as a simplified metric derived from TRP transmission and reception spectral efficiency. In link-level simulations with one TRP and one target UE, the metric is defined as shown below.
Here X denotes the number of bits correctly received by the target user, T is the simulation time, and W is the bandwidth including the required guard bands. The equations above are detailed separately for the single-numerology and mixed-numerology cases below.
Downlink Single-Numerology Case
In this case, W = BWcarrier is the total carrier bandwidth. X is the number of bits correctly received within the signal transmission bandwidth and over time in a typical fading channel. Key factors affecting X include system guard-band overhead, time-domain overhead, and BLER. For time-domain overhead, residual tails are a major concern for waveform design. The following sections discuss these factors for the single-numerology case.
System Guard-Band Overhead
The system guard band is intended to meet the baseband out-of-band (OOB) emission requirements derived from the 3GPP RF spectral mask, as shown in Figure 1(a). Guard-band calculations should be based on agreed baseband OOB assumptions. It is recommended to use the 3GPP RF spectral mask plus a margin (for example, 10 dB) as the baseband OOB requirement. The margin compensates for RF impairments including DAC quantization, power amplifier nonlinearity, and wideband noise accumulation. The system guard-band Gsys is given by the following relation.
Here Gsys denotes the width of the system guard band.
Residual Tail Length
Many filtered/windowed waveforms introduce residual tails. As shown in Figure 2, these tails can create additional time-domain overhead at DL/UL switching boundaries in TDD systems. If the tail can be accommodated by the TDD system's guard period (GP) without performance degradation, there is no extra overhead and the tail can be counted as zero. For FDD, if the tail does not introduce extra GP between adjacent subframes, it also does not incur extra overhead and can be treated as zero.
BLER Performance in Fading Channels
BLER in fading channels is used to compute the number of bits correctly received within the signal transmission bandwidth and time, excluding system guard-band and time-domain overhead. It reflects the waveform's inherent ISI/ICI impact on receiver performance, particularly considering channel multipath extensions (for example, ETU). For filtered/windowed waveforms, ISI in fading channels will be larger than in AWGN channels because convolution with filter/window responses and rich multipath will amplify inherent ISI.
Downlink Mixed-Numerology Case
If multiple subbands with different numerologies coexist within a single contiguous spectrum block, inter-subband interference arises because OFDM sine orthogonality cannot be preserved across adjacent subbands. Interference from adjacent subbands is especially severe for subcarriers at the edges of the target subband. Since users may be scheduled at edge PRBs within the target subband, evaluating performance at the edge PRB is important. Therefore, it is reasonable to evaluate spectral efficiency for a single PRB at the target subband edge.
In this case, the spectral efficiency at the edge PRB can be computed given a number of guard tones between subbands as shown below.
Here X denotes the number of bits correctly received on the edge PRB, which reflects BLER performance in a fading channel. T is the simulation time, and W is the bandwidth of one edge PRB.
Uplink Asynchronous Transmission Case
When multiple UEs are asynchronous with respect to each other, inter-UE interference occurs. Similar to the downlink case, evaluate the spectral efficiency of the interfered target UE given a number of guard tones between UEs. Power offsets between UEs at the base station should also be considered. The target UE spectral efficiency is computed as:
Here X denotes the number of bits correctly received by the target UE, reflected by BLER performance in a fading channel. T is the simulation time, and W is the bandwidth of the target UE.
