Geometry-Based Stochastic Model

Scatterers are distributed randomly within defined geometric regions (circles, ellipses, or polygons) according to specified spatial probability density functions. Unlike purely statistical models, the physical location of each scatterer is explicitly generated, enabling the calculation of Angle of Departure (AoD), Angle of Arrival (AoA), and Time of Arrival (ToA) through deterministic ray geometry. Common distributions include uniform, Gaussian, and von Mises for angular spreads.

GSCMs inherently capture the double-directional impulse response, parameterizing the channel by both departure and arrival angles alongside delay and Doppler shift. This makes them uniquely suited for MIMO system evaluation, as the spatial correlation across antenna array elements emerges naturally from the geometric configuration rather than being imposed through abstract correlation matrices.

Scatterers are grouped into clusters representing physical objects (buildings, vehicles, foliage). Each cluster contributes a set of multipath components with similar parameters:

• Intra-cluster parameters: delay spread, angular spread, and power decay profile
• Inter-cluster parameters: cluster birth-death processes and spatial consistency This structure mirrors real-world propagation measurements from channel sounding campaigns.

GSCMs support continuous time evolution by updating scatterer positions, cluster visibility, and path delays as the mobile station moves. This enables:

• Smooth transitions between Line-of-Sight (LOS) and Non-Line-of-Sight (NLOS) states
• Realistic Doppler spread generation from geometric motion
• Spatial consistency across closely-spaced user positions These features are critical for beam tracking and mobility management algorithm testing.

Major GSCM implementations include:

• 3GPP Spatial Channel Model (SCM) and its enhanced version SCME
• WINNER II and WINNER+ models, covering indoor, urban micro, urban macro, suburban, and rural scenarios
• COST 2100 model with explicit visibility regions
• 3GPP TR 38.901 for 5G NR systems from 0.5 to 100 GHz Each provides calibrated parameter tables derived from extensive measurement campaigns.

GSCMs occupy a middle ground in the modeling hierarchy:

• Higher fidelity than tapped-delay-line stochastic models due to geometric angle derivation
• Lower computational cost than full ray tracing, as only stochastically placed scatterers are evaluated
• Simplified GSCMs use single-bounce scattering assumptions, while advanced variants incorporate multiple-bounce paths and diffuse scattering components for improved accuracy at the cost of complexity.

The broader class of models to which GSCMs belong. Unlike deterministic ray tracing, a stochastic channel model describes the wireless medium using statistical distributions for parameters like delay, angle, and power. The GSCM advances this by anchoring those distributions to a specific geometric layout of scatterers, making the spatial consistency of the channel more physically realistic than purely tap-based models.

A critical spatial parameter in GSCM frameworks. Angle of Arrival (AoA) defines the direction from which a multipath component impinges on the receiver array. In a GSCM, AoA is not an abstract coefficient but is geometrically derived from the stochastic placement of scattering clusters relative to the receiver's position, enabling accurate beamforming and spatial correlation modeling.

A measure of time dispersion in a multipath channel. In a GSCM, the delay spread emerges naturally from the geometric distances between the transmitter, the stochastically placed scatterers, and the receiver. This contrasts with tapped-delay-line models where delays are assigned from a predefined power-delay profile without explicit geometric causality.

The spectral broadening caused by relative motion. GSCMs model Doppler shifts by calculating the radial velocity component between each moving scatterer and the receiver based on their geometric positions. This provides a physically consistent coupling between the Angle of Arrival and the observed Doppler frequency, a feature often missing in non-geometric statistical models.

A hybrid modeling approach that combines deterministic ray tracing for strong specular paths with a GSCM for weaker, diffuse scattering. In a quasi-deterministic channel, the dominant line-of-sight and first-order reflections are computed precisely from a 3D map, while the dense multipath tail is generated by stochastically placing diffuse scatterers around the receiver, balancing accuracy with computational efficiency.

A mathematical structure describing the correlation of fading signals across antenna elements. GSCMs inherently generate realistic spatial correlation matrices because the angular spread of arriving multipath is a direct consequence of the geometric distribution of scatterers. This makes GSCMs essential for accurately evaluating massive MIMO and beamforming performance where spatial consistency is paramount.