LOW COMPLEXITY MIMO-SCMA DETECTOR BASED ON SPHERE DECODING WITH CLUSTERED TREE ALGORITHM

• Main Author: Kurniawan, Dayat
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43450
• Institution: Institut Teknologi Bandung

In the 5th generation or 5G communication system that is expected to increase service capacity and connectivity, it becomes a challenge to deliver technological innovations that can maximize the available frequency spectrum. Sparse code multiple access (SCMA) is one of the code-based non-orthogonal multiple access (NOMA) techniques proposed to increase the capacity of 5G communication. In SCMA systems, each user has a unique code and can occupy more than one resource element (radio channel) or one resource element can be occupied by more than one users. The more users who have to be served, the level of complexity of detection on the receiving side will also increase. On the other hand, one technique to increase service capacity is to use multiple input multiple output (MIMO) techniques. By combining MIMO and SCMA called MIMO-SCMA, it is hoped that the challenges in 5G communication can be solved. However, by combining MIMO and SCMA the complexity of detection at the receiver side will increase in proportion to the number of MIMO elements used. The receiver will do the detection process twice, the first detection for MIMO and the second detection for SCMA. An algorithm innovation is needed to detect MIMO and SCMA in one operation. In this research, the detection in the MIMO-SCMA receiver is designed for low complexity where the detection process of MIMO and SCMA is done in one operation. One detection algorithm that is often used on SCMA systems is the message passing algorithm (MPA). MPA has a performance that is close to maximum-likelihood (ML) detection but has a high complexity. The main challenge in MPA is to reduce the complexity of detection with performance that is close to the detection results of the ML algorithm. The complexity of the detection of a single SCMA is proportional to the number of codewords and the number of users occupying one resource element, notated as ????????????; while for MIMO-SCMA is proportional to ????????????????????????×????????, where ???????????????????? is the number of MIMO elements used. To overcome the complexity of detection in MIMO-SCMA, this research conducts a detection design for MIMO-SCMA by combining joint factor graph (JFG) and sphere decoding (SD) with clustered tree algorithm which is called SDCT-JFG-MPA for the first time. JFG algorithm simplifies the detection of MIMO and SCMA in one operation while the SD algorithm with the addition of clustered trees reduce the computational complexity of MPA. The design is carried out on a 2x2 MIMO-SCMA system that works at 150% SCMA overloading factors with Rayleigh fading and narrowband mmWave Saleh-Valenzuela (SV) models, for indoor environments. Simulations were carried out to investigate the performance of the SDCT-JFG-MPA algorithm with various variations of the search radius (RSD) and variations of the channels depicted in the bit error rate (BER) curve and the processing time curve. The result shows that the optimum radius for single SCMA and 2x2 MIMO-SCMA with optimum BER performance is achieved when ???????????? = ????????? × 3 2 ×???????????????? and ???????????? = ????????? ×???????????????????? × 3 2 ×???????????????? respectively. Compared to JFG-MPA in 2x2 MIMO-SCMA system, SDJFG-MPA decoder brings 25% reduction of processing time, while in SDCT-JFG-MP decoder, the reduced percentage reaches 50% of JFG-MPA decoder under the same ???????? ???????? ? value.