Conception et implantation FPGA d’un émulateur de canal MIMO multitrajet pour LTE

Sujets de thèse 2013

Intitulé de la thèse
Conception et implantation FPGA d’un émulateur de canal MIMO multitrajet pour LTE
Publication du sujet sur le site de l’ABG : OUI
Nature du financement : Convention CIFRE
Domaine de compétences principal (pour l’ABG) : Sciences pour l’Ingénieur
Domaine de compétences secondaire (pour l’ABG) : Informatique, électronique
Spécialité de doctorat : Electronique des hautes fréquences, Photonique et Systèmes

Lieu de travail
Limoges/Sophia antipolis
Date Limite de candidature : 15/06/2013
Laboratoire d’accueil : XLIM/C2S2

Présentation de l’équipe de recherche
ESTE (Etude des Systèmes de Télécommunications de l’ENSIL) est un des groupes de recherche du département C2S2 de Xlim. Les activités menées au sein de ce groupe concernent l’étude des systèmes de télécommunications filaires et sans-fil selon une double approche composant et signal. Plus précisément, ces études sont associées à 3 axes de recherche étroitement liés: > L’étude d’algorithmes de traitement du signal pour les communications numériques. > L’implantation de ces algorithmes sur plateformes cibles de type DSP ou FPGA. >L’étude de l’impact des composants sur les performances système d’une chaîne de transmission numérique. Le contexte de cette thèse se situe dan s les deux premiers 1 et 2.

Résumé de la thèse en français
L’objectif de cette thèse est de concevoir et d’implanter un émulateur du canal multi trajet et multi antennes en tenant compte la corrélation spatiale entre les trajets. La cible est un FPGA Xilinx et on utilisera l’approche SoC pour profiter pleinement de toute la capacité du FPGA.

Résumé de la thèse en anglais
The goal of this PhD thesis is to design and implement a channel emulator for MIMO multipath communication channel considering the spatial correlation between paths. The target will be a Xilinx FPGA with power PC in order to exploit all its capacity.

Description complète du sujet de thèse
In order to evaluate the performance of wireless systems it is essential to test them in a real propagation environment. However, since the characteristics of the channels are time and location dependent, it becomes difficult to compare different systems fairly. That is why the engineers prefer to test their receiver system by emulating the channel rather using the unknown real channels. Exhaustive researches have been done to understand different phenomena in the propagation channels in order to obtain realistic mathematical models. These models are based on the deterministic parameters of the transmission such as distance, frequency, antenna gains, etc (site-specific wave propagation approach) 1] and also on some stochastic processes because of non-deterministic nature of the environment (statistical channel modeling approach). We focus on second approach because in this way, we bypass the physical details and generate only fading channel impulse response with accurate statistical properties that simulate real propagation channels 2][3][4].
In practice, band-pass channels can be emulated using RF circuits. However, if the carrier frequency is quite high, it is easier to work on the baseband signals. In this case, the low-pass equivalent model of the channel is used instead of the model of the channel in carrier frequencies. Today, because of the demand for high throughput systems, the bandwidth is increasing continuously and the complexity of the real time emulators are becoming so high that the use of FPGA emulator is preferred over the microprocessor or DSP designs.
In this thesis we are going to implement on FPGA a realistic propagation baseband channel for MIMO systems that comply with the LTE standard. Furthermore a frequency shift between the received signal and the local oscillator (due to Doppler Effect or oscillator limited precision) should be considered as well. The design must also be able to switch to different standard channel models according to COST 207 [5].

We have already studied the mathematical modeling and simulation of fading SISO channels in the ANR project CAMAZOC [6] where different types of channels (Gauss, Rayleigh, Rice, Nakagami) are modeled and simulated using matlab. Then these channel models have been combined to comply with COST 207 recommendations. These models and simulations can be immediately extended to MIMO in the case where different paths are independent. However, in many applications such as in the cell phones where the antennas are too close together, the paths are no longer independent. In general three types of dependence or correlation are considered between the paths: spatial correlation, time correlation and frequency correlation (Triply Selective MIMO Channels) [2]. The channel emulation taking into account these phenomena requires matrix multiplications, matrix square root calculations, and matrix Kronecker multiplications, all over complex data. To which, we should add the complexity of Doppler shift emulation and random complex numbers generation following given statistical distributions (pdf). Therefore, a real time realistic channel emulator is a real challenge that requires a smart design of the structure and resource sharing. Some efforts have been already done to implement the wireless channel. For example in [8] an FPGA architecture is proposed for generating Rayleigh and Rice channel using the method “sum-of-sinusoids”. In [7] the correlation effects have been considered for a 2×2 MIMO channel complying with WiMAX standard.
In this thesis we propose to emulate all the effects in the same circuits. The challenge is in two faces. The first one is the proper partitioning of software/hardware. It is because in the commercialized FPGAs today, microprocessors are integrated on the same chip. For example, the microprocessor integrated on Xilinx FPGAs, called power PC, can be programmed together with the logic in the FPGA parts, permitting us to have a complete System-on-Chip (SoC). The next face is related to resource sharing. Since all the channel characteristics are implemented on the same chip it is possible to share resources. The architecture we are proposing is essentially based on sequential/combinatorial coprocessors that accelerate the execution of two parallel embedded processors. The filter-based modeling approach [10] for fading channel will be considered which gives a straightforward realization suitable for FPGA design.

The candidate begins with a bibliographical phase in order to understand fading multipath MIMO channels and to familiarize himself (herself) with the laboratory equipment. At the end of the first year, a simple independent MIMO flat-fading channel emulator should be completely implemented based on the architecture for Rayleigh flat fading-channel in [9]. Then, different correlation effects are studied and simulated with fixed-format numbers. Following these simulation, the most robust method will be implemented. At the end, user friendly design together with the Human-Machine-Interface will be considered.

[1] A. Goldsmith, Wireless Communications. Cambridge, England: Cambridge University Press, 2005.
[2] C. Xiao, J. X. Wu, S.-Y. Leong, Y. R. Zheng, and K. B. Letaief, “A Discrete-time Model for Triply Selective MIMO Rayleigh Fading Channels,” IEEE Trans. Wireless Commun., vol. 3, no. 5, pp. 1678-1688, Sep. 2004.
[3] Y. R. Zheng and C. Xiao, “Simulation Models with Correct Statistical Properties for Rayleigh Fading Channels,” IEEE Trans. Commun., vol. 51, no. 6, pp. 920-928, Jun. 2003.
[4] B. E. Baddour and N. C. Beaulieu, “Accurate Simulation of Multiple Cross-correlated Rician Fading Channels,” IEEE Trans. Commun., vol. 52, no. 11, pp. 1980-1987, Nov. 2004.
[5] COST 207, “Digital land mobile radio communications”, Office for official publications of European communities, final report, Luxembourg, 1986
[6] ANR project “Caractérisation, métrologie et amélioration de la zone de couverture en mobilité” (ANR CAMAZOC) 2008-2011
[7] Fei Ren, Yahong Rosa Zheng “Hardware Emulation of Wideband Correlated Multiple-Input Multiple-Output Fading Channels” Journal of Signal Processing Systems, March 2012, Volume 66, Issue 3, pp 273-284
[8] S. Fouladi Fard, A. Alimohammad, B.F. Cockburn, « An FPGA-Based Simulator for High Path Count Rayleigh and Rician Fading, » Vehicular Technology, IEEE Transactions on , vol.59, no.6, pp.2725-2734, July 2010
[9] Alimohammad, A.; Fard, S.F.; Cockburn, B.F.; Schlegel, C., « A Compact Single-FPGA Fading-Channel Simulator, » Circuits and Systems II: Express Briefs, IEEE Transactions on , vol.55, no.1, pp.84,88, Jan. 2008
[10] Amirhossein Alimohammad, Saeed Fouladi Fard, and Bruce F. Cockburn, “Filter-Based Fading Channel Modeling,” Modelling and Simulation in Engineering, vol. 2012, Article ID 705078, 10 pages, 2012

Objectifs scientifiques de la thèse
L’objectif est de fabriquer un système qui remplace le canal MIMO multi trajet de manière réaliste en accord avec les spécifications de LTE.

Compétences à l’issue de la thèse
Le (la) candidat(e) aura une très bonne connaissance sur les systèmes de transmission numérique et les canaux de transmission. Il (elle) maîtrisera le domaine très porteur d’implantation FPGA. Les aspects théoriques sont aussi bien abordé que les aspects pratiques dans un environnement de recherche industrielle.

Mots clés (séparés par des virgules)
communication numérique; émulateur de canal; implantation numérique; FPGA;
Conditions restrictive de candidature (nationalité, âge, …) : OUI

Expérience/profil souhaité(e)
– Connaissance des systèmes de télécommunication
– connaissance d’implantation circuit numérqiue

Modalité de dépôt des candidatures
Envoyer CV et lettre de motivation au directeur de thèse.

Directeur de thèse
Adresse mail du directeur de thèse : [
Téléphone Directeur de thèse : 0555423656
Cofinancement LABEX SigmaLIM demandé : NON
Thèse pour Action transverse : NON
Fichier annexe (pdf)Projet de Recherche


Menu principal

Haut de page