To accommodate the ever-increasing mobile data traffic, it is inevitable to reduce the cell size and, as a result, increase the number of base stations. In the conventional radio access network (RAN), base stations are composed of two components; remote radio head (RRH) and base band unit (BBU). RRHs convert the radio frequency (RF) signals to the baseband electrical signals and BBUs are used to process these electrical signals. Thus, the increased number of base stations could increase the operation and installation costs of the mobile network significantly. Recently, cloud RAN (CRAN) has been proposed to solve this problem. In CRAN, all the BBUs are moved to a centralized location, while the RRHs remain at their corresponding cell sites. Thus, by using the CRAN architecture, it is possible to improve the resource utilization through the statistical multiplexing and BBU pooling, and also save the operation, maintenance, and upgrade costs of BBUs. However, this architecture requires a high-speed mobile fronthaul network connecting BBUs and RRHs. The mobile fronthaul network is typically implemented by digital optical links with common public radio interface (CPRI) or Open Base Station Architecture Initiative (OBSAI). However, the major problems in utilizing such CPRI- or OBSAI-based fronthaul networks in CRAN are the extremely high capacity and the stringent latency requirements. For example, the required data rate of the CPRI-based fronthaul optical link can be as high as 29.5 Gb/s, if long-term-evolution-advanced (LTE-A) system utilizes four component carriers (CCs) supporting an aggregated bandwidth of 80 MHz, $2 \times 2$ multiple-input multiple-output (MIMO) antennas, and 3 sectors. Also, the sub-frame processing delay in such a fronthaul network should be kept to be <1 ms, regardless of the propagation delay of the fiber.
To solve the problems associated in the mobile fronthaul based on the digital optical link, the mobile fronthaul network based on the analog optical link has been proposed. In the analog optical link, wireless signals are multiplexed in the frequency domain and transported over an optical link without format conversion. Thus, there is no need to utilize the expensive digital components (which are required for sampling the analog signal and converting it to the digital signal in the digital optical link). As a result, in the case of using the analog optical link, RRH can be implemented by using only frequency converters and bandpass filters. Also, the analog optical link can be implemented by using relatively low-speed electro-optic components due to its high spectral efficiency.
In this thesis, we experimentally evaluate the possibility of utilizing analog optical link in the next-generation mobile fronthaul network. For the cost-effectiveness, we implement this analog optical link by using a directly modulated laser (DML) and the direct-detection (DD) technique. The carrier-to-noise ratio (CNR) is used to estimate the performance of this DML-DD based analog optical link. In this link, the CNR can be limited by the thermal and shot noises generated at the receiver and the relative intensity noise (RIN) and clipping distortion occurred at the transmitter. The power spectral densities (PSDs) of thermal and shot noises depend on the temperature and received optical power, respectively. The RIN is affected by the spontaneous emission in the laser’s gain medium, optical-frequency selection at the laser facet, and the fluctuation of the laser’s injection current. The typical RIN value of distributed feedback laser diode (DFB-LD) is -150 dB/Hz. The clipping distortion depends on the root-mean-squared optical modulation index (RMS OMI). If RMS OMI is larger than 25 %, the clipping distortion can seriously limit the CNR performance of the mobile fronthaul based on analog optical link. In order to estimate the CNR performance similar to realistic condition, a realistic wireless signal should be used. For this purpose, we generate LTE-A signal used in 4G mobile network. This LTE-A signal is composed of multiple LTE signals, often referred to as component carriers (CCs), since the LTE-A system should be able to cover the LTE system (i.e. backward capability). Thus, to emulate the LTE-A signal, we combine numerous LTE signals compatible with 3GPP specifications with LTE-A signals by using the subcarrier multiplexing (SCM) technology. This SCM signal is generated by an arbitrary waveform generator (AWG). The limitation caused by the quantization effects of AWG was evaluated quantitatively to confirm the quality of the generated LTE-A signals. The quality of the SCM signal is degraded as the number of the carrier increases and the carrier frequency is high. The LTE-A signal generated by AWG is used to directly modulate the DFB-LD. The CNR and error vector magnitude (EVM) are measured to confirm the performance of the mobile fronthaul based on analog optical link. The measured data agrees well with the theoretically estimated values. From these results, we expect that 150 CCs of 20-MHz signal can be transmitted over the mobile fronthaul network implemented by using analog optical links, considering the available bandwidth of the DFB laser and its chirp. Another potential impairment source in this analog optical link is the multipath interference (MPI) induced caused by reflections in the optical fiber. In the previous reports, this MPI effect has been investigated with the quasi-static approximation. In comparison, we estimate the PSD of the MPI-induced noises without using the quasi-static approximation. Using this result, we estimate signal-to-interference ratio (SIR) required to ensure the satisfactory performance of the mobile fronthaul network implemented by using analog optical links. For example, in case when seventy-two 20-MHz LTE signals are transmitted in a fiber-optic link with link loss of 13 dB and the launched optical power is 5 dBm, the SIR should be >62 dB to achieve the CNR of 30.5 dB (which corresponds to EVM of 3 %).
계속 증가하는 모바일 데이터 트래픽을 수용하기 위하여, 셀 크기를 줄이는 것을 피할 수 없고, 결과적으로, 셀 개수를 증가시키는 것도 피할 수 없다. 기존 라디오 액세스 망에서, 기지국은 RRH (remote radio head)와 BBU (base band unit)으로 구성되어 있다. RRH는 무선신호를 처리하고, BBU는 기저대역 신호를 처리한다. 증가된 기지국의 개수는 모바일 망의 설치 및 유지 보수 비용을 증가시킨다. 최근, CRAN (cloud radio access network)가 이 문제를 해결하기 위해 제안되었다. CRAN구조에서, BBU는 중앙 기지국으로 이동 되었고, RRH만이 원래 기지국에 남아있다. 그러므로, CRAN 구조를 이용함으로써, 유저가 많은 곳에 많은 BBU를 할당할 수 있으므로 시스템이 좀 더 효율적으로 되고, BBU의 유지,보수 그리고 업그레이드 비용을 절약할 수 있다. 그러나 CRAN구조는 BBU와 RRH 사이의 링크인 프론트홀에 막대한 전송용량을 요구한다. 만약 시스템이 80 MHz의 대역폭, 2 by 2 MIMO, 그리고 3 섹터를 사용한다면, 프론트홀의 전송용량은 29 Gb/s까지 증가한다.
이러한 문제를 해결하기 위해 아날로그 광링크가 제안되었다. 아날로그 광링크에서, 무선 신호는 주파수 영역에서 다중화 된 뒤 그리고 포맷 변환 없이 광링크를 통하여 전송된다. 그러므로, 비싼 디지털 소자가 사용될 필요가 없다(아날로그 디지털 컨버터 및 디지털 아날로그 컨버터). 결과적으로, 아날로그 광링크를 사용하는 경우, RRH는 주파수 변환기와 대역통과필터로 구성 될 수 있다. 또한 아날로그 광링크는 높은 대역 효율을 가지므로 상대적으로 저속의 광/전소자를 이용하여 구현이 가능하다.
