Under the Hardware design section, as part of the progress made in the hardware design for C-RAN, different test beds are studied to evaluate the merits and demerits involved in different functional split options.
There were already some existing works , , , , which discuss various technological advancements made in C-RAN. An introductory to C-RAN is given in , which explains a high-level overview of C-RAN architecture, network virtualization, and transport network. In , fronthaul design aspects, as well as the compression techniques, are discussed in detail. The field trials and testbed development for C-RAN are also highlighted. But the state-of-the-art technologies discussed in this work are not exhaustive and up to date. A comprehensive survey on various technologies, architecture, challenges, requirements, and proper potential solutions to achieve an efficient C-RAN fronthaul is studied in . Schemes that can help in reducing system complexity, bandwidth requirement, cost, and latency are considered as part of this work. To reduce the overhead of the fronthaul, different functional splits proposed from both academia and industry are highlighted in . The investigation of this work is about, how the choice of functional split affects the fronthaul network, both in terms of latency and bitrates. The comprehensive literature ,  is dedicated to a focused area of C-RAN architecture. None of the research, except , are discussing the fronthaul design aspects of C-RAN, which is the primary concentration area of our work.
The rest of the paper is organized as follows. section 2 presents a literature survey on C-RAN systems architecture. The advantages of C-RAN are listed in section 3. section 4 describes the different types of fronthaul solutions with their pros and cons. The technical challenges faced by the present fronthaul types are presented in section 5. In section 6, various I/Q sample compression techniques are studied under Fronthaul compression. All compression schemes, such as distributed uplink compression, multivariate downlink compression, and point-to-point compression, are thoroughly discussed. The existing literature available on different C-RAN testbeds is presented in section 7. Our testbed, which is based on the Labview framework, is also shown in the same section. Finally, in section 8, the summary of the whole literature is provided.
2 C-RAN Architecture
From the inception phase of C-RAN, there has been a tremendous change in architecture, and this is evolving continuously. At the same time, a significant part of work is dedicated to the evaluation of architecture. To combat the current situation with data proliferation, the initial architecture is not fit for now, and it may so happen, existing architecture will not be fit for the near future. Thus, detailed and thorough surveying of this topic will help us to know the need for the continuous architectural change and future scope as well.
Though C-RAN architecture evaluation is the most dynamic field and changes a lot from the initial time, significant changes have been happening since 2015. But to make this survey more comprehensive, we will discuss some earlier work as well.
Handover is always a concern for cellular networks. In , the handover performance gain of C-RAN over traditional RAN for GSM, UMTS is discussed. Due to the synchronous nature of handovers in C-RAN, the performance gain is tenfold than that of decentralized RAN (D-RAN). Because of intrapool softer handover in C-RAN, the transport bandwidth requirement is also reduced. As the BBU pool is centralized and many RRHs are connected to the same BBU, handover management is much easier than a decentralized one.
Whether it is centralized RAN or distributed RAN, load balancing and massive data processing have always been a research topic. In the case of the traditional BS scenario, the processing element in each BS is designed to handle the peak load. So, the general processing unit does not work in this case. Most of the time, those resources are idle. But thanks to cloud computing, its related resource management, and the virtualization technique, any time resource can be added or removed. In , load analysis of C-RAN has been done. Apart from that analysis, a GPP based C-RAN architecture also proposed for the improvement of scalability, computation cost, consumption of power, unnecessary data movement, and inter-server communication. This work also maximizes CPU strength and flexibility. But for strengthening CPU power, some task scheduling algorithm is also viable, but that was not discussed. The mechanism of inter-server and intra-server communication is also not mentioned.
In , a novel end-to-end transport network solution is proposed. In this work, an innovative, WDM-PON based solution is discussed, where the availability of dedicated fiber is limited. A converged metro access network architecture has been proposed to support both centralized and distributed radio baseband solution. But this is mainly a backhaul related solution.