Graduate School and Research Center In communication systems

Imran LATIF

Imran LATIF
Imran LATIF
Eurecom - Mobile Communications 
Phd Student ( 2010 - 2013)
Alumni EURECOM

Thesis

Scalable System Level Evaluations for LTE Using PHY Abstraction

Responsible(s)

 

The main focus of this thesis is to highlight the importance of PHY abstraction for the system level evaluations in the framework of 3GPP Long Term Evolution (LTE) networks. This thesis presents a pragmatic approach towards the use of PHY abstraction in LTE based system level simulators. PHY abstraction is an extremely valuable low complexity tool for efficient and realistic large scale system evaluations. This thesis shows that apart from the primary purpose of PHY abstraction of providing instantaneous link quality indicator for the purpose of system level evaluations, it can be further used for an improved channel quality indicator (CQI) feedback based on the different antenna configurations and for the performance prediction of LTE networks based on the real life channel measurements.

 

This thesis is mainly divided into two parts; methodologies and applications. The first part presents the complete design and validation methodology of PHY abstraction schemes for various antenna configurations corresponding to different transmission modes in LTE. The validation is performed using link level simulators and it also highlights the calibration issues necessary for the PHY abstraction to be accurate in predicting the performance of capacity achieving turbo codes.

 

For the special case of multi-user multiple-input multiple-output (MU

MIMO) an improved PHY abstraction model is presented which exploits the knowledge about the fact that multi-user interference (MUI) deviates greatly from its Gaussian assumption and therefore must be exploited while predicting the PHY performance especially in the case of interference aware receivers. The first part also presents a novel semi-analytical PHY abstraction approach towards incorporating the incremental-redundancy hybrid automatic repeat request (IR HARQ) for a wide variety of resource block assignment in LTE and it further reduces the storage requirements for PHY abstraction by bringing down the number of required reference curves to only three from hundreds.

 

The second part presents the applicable scenarios where the concepts of PHY abstraction can be effectively utilized. The most important and primary use of PHY abstraction is to reduce the necessary simulation time in system level simulators by predicting the link quality based on the instantaneous channel by a fewer number of operations than are required for the processing of complete PHY procedures. Thus by doing so it was found that the overall simulation time for large scale simulation can be decreased enormously by speeding up the simulation by a factor of at least 30 times. Which means that the simulation which took the run time of a month can now be performed with in a day only. Furthermore, it is shown in this part that the performance predicted by PHY abstraction matches the performance obtained by running the simulations with full PHY procedures.

 

The second part also presents the complete methodology for the use of PHY abstraction in the CQI feedback calculation necessary for the scheduling purposes. In this part we propose to exploit the knowledge on the antenna configuration for the CQI mapping for an improvement in the link adaptation at the MAC layer. Last but not the least, the second part presents a complete methodology for the PHY abstraction to be used for the validation of different LTE transmission modes based on real channel measurements in rural areas at 800 MHz. We use concepts from PHY abstraction to predict the performance of LTE network by means of the channel measurements which were stored during a measurement campaign. We showed that using PHY abstraction not only the performance of the system being operated in different transmission modes can be predicted but also it can be used for the validation of soft LTE modems. This shows that PHY abstraction models has lots of potential and are an important tool for the research community in both industry and academia.

 

All of the results presented in this thesis have been obtained using open-source OpenAirInterface platform which is implemented in highly optimized C. It is one of the very few open-source platforms who have the capability of performing simulations with both full PHY or PHY abstraction and are composed of both link level simulator and system level simulator. Further it can be used for simulation, emulation and real life demonstrations using the LTE hardware especially designed for it.