3GPP CDL

CDL models represent the wireless channel as a finite set of discrete clusters, each composed of 20 rays with identical delay but slightly varying angles. Each cluster is defined by its normalized delay, power, and angular parameters for both departure (AoD, ZoD) and arrival (AoA, ZoA). This geometric approach captures the spatial consistency essential for evaluating beamforming and massive MIMO algorithms.

3GPP TR 38.901 defines three CDL profiles for different propagation environments:

• CDL-A: Non-line-of-sight (NLOS) with moderate angular spread, representing urban macro cells
• CDL-B: NLOS with larger delay spread, modeling bad urban or hilly terrain
• CDL-C: Line-of-sight (LOS) with a dominant direct path and weaker clusters Each profile specifies per-cluster delays, powers, and angular spreads calibrated from extensive channel measurement campaigns.

CDL models are inherently scalable to any desired bandwidth. The standardized delay values are normalized, and the model applies delay scaling to map them to the target sampling rate. For 5G NR simulations, the cluster delays are scaled to match the OFDM numerology and subcarrier spacing, ensuring the channel impulse response aligns precisely with the FFT window and cyclic prefix duration.

Each CDL cluster is characterized by angular spreads—the per-cluster Angle Spread of Departure (ASD) and Angle Spread of Arrival (ASA). These spreads control the spatial correlation properties of the channel, directly impacting MIMO multiplexing gain and beamforming resolution. Clusters with narrow angular spreads produce highly correlated antenna responses, while wide spreads create rich scattering environments favorable for spatial multiplexing.

CDL models couple with arbitrary antenna array geometries through the steering vector calculation. The model computes the complex channel coefficient for each antenna element by summing the contributions of all rays across all clusters, weighted by the element's phase response at the specified angle of arrival. This enables accurate simulation of polarized arrays, cross-polarization discrimination (XPD) , and dual-polarized panel arrays used in 5G NR base stations.

CDL is the primary link-level channel model in 3GPP 5G NR evaluations, complementing the Tapped Delay Line (TDL) model for simpler scenarios. It is used to benchmark channel estimation algorithms, CSI feedback schemes, and precoding performance under standardized propagation conditions. Unlike system-level models such as the Urban Macro (UMa) pathloss model, CDL focuses on the detailed spatial and temporal structure of the channel impulse response.