부산대학교 도서관

Since South Korea and United States initialize the commercial operation for 5G New Radio (NR), our country in turn release the licenses of 5G NR in January 2020 and then start the commercial operation at second half of year 2020. Moreover, the Executive Yuan have launched a four-year plan of 20 billion dollars to develop critical 5G NR technique and 5G NR system test platform. In doing so, our country can be expected to be the key partner of the 5G-NR supply chain in the world. In this thesis, we present the Simulink-Zedboard software-defined radio (SDR) hardware implementation for the 5G NR cell search procedure with efficient finite-length impulse response (FIR) filter architecture. In the design, we follow the 3GPP-v16 working standard for 5G NR [1] and employ the cell search algorithm with low computing complexity and high accuracy [2]. The primary characterizations of the algorithm in [2] are stated as follows: (1) The high-accuracy time-average timing-frequency estimation is employed to enhance the cell index detection rate greatly, and (2) The high cell index detection rate up to 70% can still be accomplished even though under the extremely low signal-to-noise ratio (SNR) of SNR = -6 dB and severe channel environment with ETU300 channel model. In doing so, the commercial requirements can be achieved. In the thesis, we employ the co-simulation design method with Simulink-Zedboard SDR platform quickly to implement the prototypes of the important devices of the 5G-NR cell search procedure with efficient finite-impulse-response (FIR) architecture. In the design of the 5G-NR cell search procedure algorithm, the flowchart is divided into four parts. First, estimate the coarse symbol timing and the fractional carrier frequency offset (FCFO) by using the time-averaged time-frequency parameters estimation algorithm. Second, detect the starting position of the 5G-NR Primary Synchronization signal (PSS). Third, detect the integer CFO (ICFO) and the sector cell index (S-CID) simultaneously. Finally, detect the group cell identity (G-CID). In the implementation, we use the algorithm proposed in [16] and [17] as a basis to design the cell search procedure with the highly efficient FIR filter architecture. In this thesis, we present three different types of the highly efficient FIR architecture types, including Type-I, Type-II, and Type-III. The three FIR types, which are discussed and compared in Chapter 4. Besides, the transversal FIR filter architecture is considered for fair comparison. Finally, the co-simulation design results of Zedboard hardware design and the software Simulink design are consistent well with each other. It is showed that the hardware design is correct and stable.