Estimation of MIMO Channel Using Suboptimal Particle Filtering

Particle filters have been successfully employed to track multiple-input multiple-output (MIMO) fiat fading channels for mobile wireless communications. However, an optimal importance density cannot be always found to optimize the performance of a particle filter. A suboptimal importance density such as the prior can be used; but it has a problem of ignoring the current observations. In addition, like other Bayesian filtering techniques, particle filters require the knowledge about dynamic noise and measurement noise to approximate the posterior distribution. This paper presents a suboptimal particle filter that overcomes the problems of uncertain noise variance, and the drawback of the prior importance density. Computer simulation of a 2 times 2 MIMO system is presented to illustrate the performance of the proposed particle filter technique.