A stochastic MIMO model utilising spatial dimensionality and modes

Abstract

This paper presents an efficiently parametrised model for second-order-statistics dominated MIMO channels. Recently, new MIMO models have been developed to emulate the statistics of real measurements: (1) analytic models which parametrise the statistics of the channel gains, and (2) geometric models which interpret the channel as separate multi-paths. Unfortunately analytic models are tied to the measurement array geometry, while geometric models significantly increase model complexity. We present a new stochastic framework, based on a modal decomposition of the MIMO channel, which allows channel models for arbitrary array geometries from a single set of measured data. Such a framework yields simple MIMO models that efficiently parametrise the channel, with adjustable accuracy. Results show that the new models match the capacity of real and simulated data as well as similar models.