Design of Printed Coprime Antenna Arrays for Direction Finding Applications. Masters thesis, King Fahd University of Petroleum and Minerals.

Preview

PDF MS_Thesis_Complete.pdf - Published Version Download (10MB) | Preview

Arabic Abstract

يعتبر تقدير الاتجاه الذي وصلت عن طريقه الموجة الكهرمغناطيسية من المكونات المهمة للجيل الخامس من الاتصالات اللاسلكية حيث ترسل المحطات الأرضية إشاراتها إلى كل مستخدم على حدة مما يزيد من السعة الكلية للنظام اللاسلكي. ويحتاج تقدير الاتجاه إلى مصفوفات من الهوائيات لاستقبال الإشارة وتحليلها لمعرفة الزاوية التي سقطت منها الأشعة الكهرومغناطيسية. في السابق كان الاعتماد على المصفوفات المنتظمة لهذا الغرض حيث تكون المسافة بين الهوائيات ثابتة. إلا أنه تبين أنه من الممكن الحصول على نتائج أفضل في تحديد الاتجاه عند استخدام مصفوفات غير منتظمة من الهوائيات حيث لا يشترط أن تكون المسافة بينها ثابتة. تحديدا، أثبتت الأبحاث أن استخدام مصفوفات غير منتظمة من الهوائيات يزيدعدد الإشارات التي يستطيع النظام تحديد اتجاهها في الوقت نفسه. من هذه المصفوفات ما يسمى بالمصفوفات الأولية نسبيا أو coprime arrays. بعد مراجعة الدراسات التي بحثت في المصفوفات غير المنتظمة تبين أن معظمها اعتمد على هوائيات مثالية وركز دراسته وبحثه على الخوارزميات مع إهمال تأثير الهوائي. في بحثنا هذا، صممنا مجموعة من مصفوفات الهوائيات غير المنتظمة وأدخلناها في الخوارزميات لتقدير الاتجاه الذي تسقط منه الموجات الكهرومغناطيسية فاستطعنا دراسة تأثير خواصّ الهوائي على تحديد الاتجاه. الهوائيات المصممة في هذا البحث اشتملت على نوعين يسميان ال patch antennas وال monopole antennas واستهدفت الأجهزة المحمولة وأجهزة الحاسوب اللوحي. استخدمت في هذا البحث برامج لمحاكاة خوارزميات تحديد الاتجاه بعد إدخال خواص الهوائيات المصنعة لأجل هذا الغرض، كما وأجريت تجارب عملية لإثبات النتائج التي وصلنا إليها عن طريق المحاكاة وفي النهاية حددنا مقدار الخطأ الناتج عن استخدام هوائيات فعلية بدل الهوائيات المثالية الموجودة في معظم الدراسات.

English Abstract

In fifth generation (5G) wireless communication systems, beamforming will play a major role in increasing the overall capacity by pointing individual beams from the base station to the users and vice versa. For such a system to work, both the base station and the mobile user need to have the capability to estimate the Direction of Arrival (DOA) of the incoming electromagnetic waves. Uniform linear antenna arrays (ULAs) are usually used for DOA estimation with the help of various algorithms. It was shown recently that nonuniform arrays can provide comparable DOA estimation performance with a smaller number of antennas. Most of previous work on DOA estimation focused on improving the utilized algorithms and overlooked the effects of hardware parts, i.e. antennas, on the algorithm performance. Only few works incorporated basic antenna behavior in the algorithms while most of the remaining literature assumed the antennas to be isotropic radiators. In this work, we highlight this gap in previous works and provide several additions to address such shortcoming. Microstrip antenna arrays are designed based on the coprime array (CPA) concept. The first array is a patch-based 4-element CPA operating in the 5.8 GHz band, intended to be used in handheld wireless devices. This array demonstrated the ability to use CPAs in small form factor terminals, something that was not studied enough in the literature, as most of the work demonstrated the advantage of large sparse arrays over ULAs, leaving the question of “how applicable this is to small arrays” open. The second and third arrays target tablet-based devices and consist of four printed patch and monopole antennas respectively, arranged in a coprime configuration and operating in the 2.1 GHz band. The complex radiation patterns of the arrays are incorporated in the DOA estimation simulation and the results are compared with those obtained by using isotropic antennas to quantify the error introduced when using physical antennas. With the 5.8 GHz patches CPA, assuming 4 sources impinging the array from angles falling within -50° and 50°, the average error between the isotropic and physical cases was 0.51°. Beyond this range of angles, the error can be as high as 7.6° assuming only one source at an angle of -80°. In the 2.1 GHz band, the average error between isotropic and patch cases for four sources impinging the array within the -50° and 50° range was 0.34° only, rising to 14.89° with one source located at -70°. The monopole arrays were found to introduce higher error compared to the patch case. An average error of 0.62° was found with four sources located within the -50° and 50° range. With only one source located near -90°, the error was 5.25°. The possibility to reduce the inter-element spacing of the CPA is investigated in the 5.8 GHz band, where we obtained 2.6° estimation error in DOA using 0.31λ. To study the benefits of coprime configuration over ULAs, 4-element and 5-element ULAs consisting of patch-based and monopole-based elements, operating in the 2.1 GHz band, are also designed, fabricated and used in DOA simulations. A typical 4-element ULA of patches can detect the DOA of three sources with a root mean square error (RMSE) of 0.79° at 20 dB SNR. However, by re-arranging the 4-elements according to the coprime concept, four sources can be detected with 0.75° RMSE at the same SNR. To be able to detect 4 sources using ULAs, a 5-element ULA will be needed, and the RMSE will be 1.08°. A monopole-based 4-element ULA can detect 3 simultaneous sources with 7.87° RMSE at 20 dB SNR. A CPA of monopoles can detect 4 sources with an RMSE of 0.64°, while a 5-element ULA can detect 4 sources with 1.19° RMSE at 20 dB SNR. DOA estimation experiments are conducted on the six arrays operating at 2.1 GHz using a software defined radio (SDR) setup in a laboratory environment and the results are compared with the simulation. With a single source located at varying DOA angles, the average difference in RMSE between simulation and experiments for the patch-based CPA was 2.87°, while for the monopole-based CPA, it was 1.23°. Experiments using two sources were also carried out, and the results were compared with those obtained by running equivalent simulations. For the patch-based CPAs, the difference in error between the simulation and the experiment was only 1.4°. However, due to multipath and lower polarization purity in monopole elements, the monopole experiments showed extra 2.63° error over the simulation predictions.

Item Type:	Thesis (Masters)
Subjects:	Electrical
Department:	College of Engineering and Physics > Electrical Engineering
Committee Advisor:	Sharawi, Mohammad
Committee Members:	Muqaibel, Ali and Nasir, Ali
Depositing User:	AHMAD OWEIS (g201601160)
Date Deposited:	31 Dec 2018 12:43
Last Modified:	31 Dec 2020 07:26
URI:	http://eprints.kfupm.edu.sa/id/eprint/140872