Optimasi Sistem Komunikasi Milimeter Wave MIMO dengan Coordinated Multipoint dan Reconfigurable Intelligent Suraface Berbasis Deep Reinforcement Learning

Ghozali, Victorio E. C. Muhamad Al (2025) Optimasi Sistem Komunikasi Milimeter Wave MIMO dengan Coordinated Multipoint dan Reconfigurable Intelligent Suraface Berbasis Deep Reinforcement Learning. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022231074-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Peningkatan permintaan dan variasi pengguna menimbulkan tantangan signifikan bagi jaringan seluler konvensional. Salah satu solusi adalah dengan menambahkan base station (BS) berdaya rendah, tetapi ini dapat menyebabkan interferensi antar sel (ICI). Teknik Coordinated Multi-Point (CoMP) telah diteliti bisa untuk mengatasi ICI dengan berbagi informasi saluran antar BS melalui tautan backhaul berkecepatan tinggi memanfaatkan teknik beamforming yang tepat. Namun, dalam komunikasi mmWave yang rentan terhadap hambatan, CoMP dengan beamforming tidak selalu menjamin keandalan jaringan. Oleh karena itu, muncul konsep Reconfigurable Intelligent Surfaces (RIS) yang dapat meningkatkan kapasitas dan cakupan sistem dengan mengarahkan phase shift ke arah yang diinginkan tanpa memerlukan daya tambahan untuk pemrosesan sinyal sekaligus mengubah hambatan berupa pantulan yang fungsional ini dipertimbangkan dalam membantunya. Penelitian ini mengusulkan optimasi dengan memanfaatkan teknik Deep Reinforcement Learning (DRL) untuk optimasi RIS dan Block Diagonalization (BD) untuk bagian precoding dalam sistem RIS-CoMP-MU-MIMO dengan mempertimbangkan lokasi dan efek lingkungan tropis untuk memaksimalkan kapasitas kanal dan sumrate. Masalah optimasi ini kompleks karena melibatkan beamforming aktif dari BS dan beamforming pasif dari RIS sehingga dilakukan Alternating Optimization (AO) hingga mendapat hasil konvergen. Hasil penelitian membuktikan bahwa dengan perbandingan dari penggunaan CoMP, CoMP-RIS dengan phase shift random, dan CoMP-RIS dengan phase shift optimal membuktikan bahwa hasil menunjukkan peningkatan signifikan dalam beberapa lokasi maupun efek lingkungan yang ada dalam performa komunikasi kapasitas kanal dan sumrate.
===================================================================================================================================
The increasing demand and diversity of users pose significant challenges for conventional cellular networks. One potential solution is to add low-power base stations (BS), but this can lead to inter-cell interference (ICI). Coordinated Multi-Point (CoMP) techniques have been explored to mitigate ICI by enabling BSs to share channel information via high-speed backhaul links and utilizing appropriate beamforming strategies. However, in millimeter-wave (mmWave) communications, which are highly susceptible to blockages, CoMP with beamforming does not always guarantee network reliability. To address this, the concept of Reconfigurable Intelligent Surfaces (RIS) has emerged, which can enhance system capacity and coverage by directing phase shifts toward desired directions without requiring additional power for signal processing. RIS can transform environmental reflections into functional components that assist communication. This research proposes an optimization framework that leverages Deep Reinforcement Learning (DRL) for RIS configuration and Block Diagonalization (BD) for precoding in a RIS-CoMP-MU-MIMO system, considering user location and tropical environmental effects to maximize channel capacity and sum rate. The optimization problem is complex due to the joint involvement of active beamforming at the BS and passive beamforming at the RIS. Therefore, an Alternating Optimization (AO) approach is employed until convergence is achieved. The results demonstrate that, compared to standard CoMP and CoMP-RIS with random phase shifts, the proposed CoMP-RIS system with optimally tuned phase shifts shows significant performance improvements in terms of channel capacity and sum rate across various locations and environmental conditions.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	CoMP, DRL, mmwave, MU-MIMO, RIS
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5103.2 Wireless communication systems. Two way wireless communication T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Victorio E. C. Muhamad Al Ghozali
Date Deposited:	23 Jul 2025 02:22
Last Modified:	23 Jul 2025 02:22
URI:	http://repository.its.ac.id/id/eprint/120665

Actions (login required)

View Item