Dear Dr Haider

I'll provide an answer that is valid for outdoor environments and frequency non-selective channels.

For NLOS channels, the most common statistical model is the Rayleigh fading. Several reasons lead the communication community to adopt this model. First, it represents the worst fading that signals can encounter. Second, it is mathematically tractable when analyzing the system performance. Third, it can be easily generated using simulators. Some other fading models can be used such as Nakagami-m, Weibull, and \alpha-\mu distributions, etc but they can fit certain channel environments and their tractability can be limited. It is interesting to know that Rayleigh fading can be derived from the fading models I mentioned through setting their parameters.

For LOS channels, Rician fading model is widely used. 

I think it is important also to know how to accurately simulate Rayleigh (Rician) fading in mobile channels. There are many simulators such as filtered white Gaussian noise (FWGN), Clarke, Jake, etc. Please have a look over the below paper which provides a recent simulation models:

http://ieeexplore.ieee.org/document/4027602/

Regards 

For outdoor channels, I suggest consulting ITU-R P.1411.

It would be helpful if you could formulate your question more precisely, Haider. There are many parameters: indoor/outdoor/frequency/distance/narrowband/wideband/spatial channel ????

Do you want path loss models, fading distributions, etc. etc.

Then type of environment and scale of its features.

The most accurate  model for a particular environment is likely to be derived from measurements taken in similar conditions and environment.

Generally, you would expect Rician in LOS and Rayleigh in NLOS multipath environments.

If we consider LOS models for the fading channel then Rician distributions would be the most suitable one as in case of satellite links.

However for NLOS case for an outdoor channel environment, Rayleigh fading (for the amplitude variation) as well m-Nakagami distributions are the best options. 

In case of NLOS for an indoor environment you should think about Saleh-Valenzula distribution which would also cater for the time dependent variations.

Regards

Dear all, many thanks for your answers.

For  LOS communication  Rician distributions  and for NLOS case Rayleigh fading channel would be the most suitable.