Multiple-Input Multiple-Output Channel Models: Theory and Practice

4

CLUSTER MODELS

One of the major drawbacks of many correlative channel models is their inability to model time-variation. Some channel statistics, such as received power, can change dramatically with small variations in transmitter/receiver position. Thus, it is important, especially in a mobile environment, to model the time-variant nature of the channel.

Cluster models bridge the gap between correlative models, which are largely stochastic in nature, and ray-tracing models, which rely more heavily on a deterministic geometry. Physically, clusters are associated with groups of scatterers in the channel. The energy reflected from these scatterers is located at different points in space. This leads to energy arriving at different angles of arrival (AoAs) and angles of departure (AoDs), as viewed from the receiver or transmitter, respectively (78). The different path lengths also lead to different delays. The size of the cluster is determined by its spread, as seen from both link-ends. The cluster can be broken down into a number of multipath components, each with its own AoA, AoD, and power. Cluster modeling reduces the behavior of clusters to a set of stochastic parameters. For most cluster models, the parameters are measured from real environments. Thus, cluster models reflect real-life channels in that their parameters are taken from measurements.

More recently, cluster models have focused on time-variations due to movement. The movement may be at the receiver, transmitter, ...