At the time that this special issue was in preparation (during 2016) we were witnessing a major research and development (R&D) effort around the world centered on the technologies that will constitute the foundation of the next generation of core and access networks supporting mobile users. Several projects were initiated worldwide in order to support the standardization of the fifth generation (5G) of cellular systems, and all major telecom players were racing toward the first commercial deployments that are estimated to take place between 2018 and 2020.

Future 5G networks will give rise to a wide range of new services with extreme requirements (in terms of data rates, latency, reliability, energy efficiency, etc.), such as ultrahigh-definition (UHD) video streaming, augmented and virtual reality, cloud gaming, smart homes, connected cars, remote control of machines, etc. As a result, the 5G networks need to support unprecedented requirements for the wireless access connection to the end-user devices, targeting cell throughput capacities in the range of 10–100 Gb/s and peak access rates per user (or even per connected device) on the order of 1 Gb/s, while maintaining low latency targets to support real-time services. Certainly it will be extremely challenging for network designers to develop the solutions to support such new network targets and service requirements for fast moving mobile users.

For the evolution toward 5G, it is envisioned that optical networking will play a major role in supporting the requirements, while reducing the deployment costs through the introduction of novel converged fiber wireless (FiWi) networks with fundamentally new PHY layer architecture, data link and network layer functionalities, and interfaces with the service layer. The target for the optical network is to support the distribution and collection of millimeter-wave radio signals, enabling the greatest flexibility for the accommodation of the wireless network requirements, while reducing the cost of the wireless access points (i.e., remote radio heads/remote antenna units). The overall network design should take into consideration the control and monitoring functionalities and capabilities of the network, and is envisioned to be based on software defined network (SDN) concepts and advanced virtualization capabilities with the use of an elastic flexible transport infrastructure and network function virtualization (NFV) elements.

With this special issue of the Journal of Optical Communications and Networking (JOCN) we tried to attract contributed papers that discuss the relevant technology solutions, together with several invited articles targeted to specific R&D efforts so as to provide complete coverage of the 5G topics for which optical networking can play a key role. The topics addressed collectively by the articles that were accepted for publication in this special issue include optical access network solutions supporting fronthaul, cloud-RAN architectures, energy efficiency, optimized network design, converged optical-mobile system solutions, and flexible optical networking. In the following paragraphs we present a brief overview of the key areas and main findings discussed in each one of the special issue articles, starting with the invited contributions.

In the invited paper “Emerging Optical Access Network Technologies for 5G Wireless” by Xiang Liu and Frank Effenberger of Futurewei Technologies, Huawei R&D, USA, the authors review emerging optical access network technologies that aim to support 5G wireless networks with high capacity, low latency, and low cost and power consumption per bit, with emphasis on advanced modulation and detection techniques, digital signal processing (DSP) tailored for optical access networks, the efficient mobile fronthaul (EMF) technique, and the need for coordination between radio access networks (RANs) and passive optical networks (PONs). In the article there is a nice example that shows how demanding are the fronthaul capacity requirements of emerging 5G networks: for a hypothetical 5G mobile network deployment scenario utilizing 200 MHz carrier aggregated signals, $64\times 64$ M-MIMO and 3 directional sector antennas, there is a requirement for 240 10 Gbps Common Public Radio Interface (CPRI) (option 7) fronthauling interfaces to connect the remote radio units (RRUs at the antenna site) with the centralized baseband units (BBUs) in cloud radio-access networks (C-RANs). In other words, the total required fronthaul data capacity reaches the staggering value of 2.4 Tb/s. To realize such capacities in a cost-effective way, we certainly need advanced optical communication systems and new techniques to support fronthauling as discussed in other articles in this special issue.

The invited paper by Shinobu Nanba, Naoki Agata, Naoya Nishi, and Kosuke Nishimura, “Geographical Constraint-Based Optimization Framework for C-RAN Design,” contributed from the KDDI R&D Laboratories, addresses an issue of network design optimization in deploying C-RAN for 5G fronthaul networking. A C-RAN network can take advantage of pooling BBU resources for fronthaul services within a cluster of multiple remote radio heads (RRHs) at a central office (CO), which leads to an interesting optimization problem to find a minimum-cost cluster set under constraints of radio interferences among RRHs and radio latency. C-RAN clusters are capable of controlling radio transmissions of RRHs from a centralized BBU to minimize radio interference, and sharing the BBU resources among RRHs of the cluster hence can enhance areal spectral efficiency and reduce deployment cost. In a large area, how to design clustering under geometrical constraints becomes critical for cost effective 5G mobile networking. This paper proposes a suboptimal but efficient cluster design algorithm for scalable 5G mobile networking.

In the paper “Dynamic Virtual Network Connectivity Services to Support C-RAN Backhauling,” Adrian Asensio, Marc Ruiz, Luis M. Contreras, and Luis Velasco from the Optical Communications Group (GCO) at Universitat Politècnica de Catalunya (UPC), Barcelona, Spain, and Telefónica Investigación y Desarrollo (TID), Madrid, Spain, discuss the evolution of RANs from distributed to centralized architectures, with emphasis on the possible options and trends for realizing the C-RAN architecture based on different functional splits, while they address the issue of how to better support dynamic mobile services that have strict constraints. They propose a dynamic customer virtual network (CVN) reconfiguration, where service requests must include quality of service constraints to ensure specific delay constraints, as well as bitrate guarantees, which are considered by the underlying Multiprotocol Label Switching (MPLS) virtual network that supports the creation and releasing of connections. Their proposal is evaluated based on exhaustive performance evaluation analysis using simulations, and it is found that the delay and the hardware complexity can be noticeably reduced when the MPLS network is dynamically managed in comparison to a static MPLS alternative.

The article “Switched Ethernet Fronthaul Architecture for Cloud-Radio Access Networks” by Philippos Assimakopoulos, Mohamad Kenan Al-Hares, and Nathan J. Gomes of the Communications Research Group, University of Kent, Canterbury, UK, discusses a fronthaul system that is based on the transport of sampled radio signals from/to base station BBUs to/from RRHs. The design considers a pure-Ethernet switched architecture that uses virtual local area network (VLAN) identifiers for the RRHs and flow identifiers for the antenna ports, and is compatible with current standardization definitions. A comprehensive analysis of the latency constraints imposed by the use of Ethernet in the C-RAN fronthaul transporting mixed traffic has been carried out. Furthermore, the performance was evaluated based on a low-cost testbed that uses “smart SFP” in-line probes to obtain measurements from an Ethernet fronthaul transporting mixed traffic. The measurements showcased how background traffic affects the hybrid-automatic repeat request (HARQ) retransmissions. The authors conclude that for an efficient realization of their system, the Ethernet switch scheduler would be a key component of the fronthaul and has to be considered when making capacity predictions.

In the manuscript “Future Proof Optical Network Infrastructure for 5G Transport” by Paola Iovanna, Fabio Cavaliere, Francesco Testa, Stefano Stracca, Giulio Bottari, Filippo Ponzini, Alberto Bianchi, and Roberto Sabella of Ericsson, Pisa, Italy, the authors present a novel transport architecture able to serve as backhaul and fronthaul and which is used to convey radio traffic over the same optical infrastructure. The proposed solution is based on advanced photonic technology (e.g., optical switches and multiwavelength transceivers implemented using photonic integration) that is used to provide optical connectivity complemented with a dedicated agnostic framing, a deterministic switching module, a flexible control paradigm based on a layered scheme and on the slicing concept to facilitate optimal interaction of transport and radio resources while preserving a well-demarcated mutual independence. The performance of the proposed transport architecture is evaluated using both simulations and experiments that also demonstrate the targeted features.

The article “Quasi-Passive Optical Infrastructure for Future 5G Wireless Networks: Pros and Cons” by Apurva S. Gowda, Leonid G. Kazovsky, Ke Wang, and David Larrabeiti from the Department of Electrical Engineering at Stanford University, the University of Melbourne, Australia, and the Telemetica Engineering Department at Universidad Carlos III de Madrid, Spain, studies the applicability of a special type of quasi-passive wavelength-selective switch [called the quasi-passive reconfigurable (QPAR) device] to implement reconfigurable backhaul in 5G fronthaul networks. The functionality, scalability and performance of the QPAR node [and variations like the pseudo-passive reconfigurable node (PPAR)] are discussed, and the performance of a QPAR-based network architectures is evaluated. It is found that the QPAR node in a hierarchical network can reduce the average latency while extending the reach and quality of service of the network.

In the manuscript “Optical Versus Electronic Packet Switching in Delay-Sensitive 5G Networks: Myths Versus Advantage,” authors Pablo Jesus Argibay-Losada, Yuki Yoshida, Akihiro Maruta, and Ken-ichi Kitayama from Osaka University, Osaka, Japan, demonstrate based on a justified analysis that optical packet switching (OPS) implemented with bufferless switches operating in medium and high-load regimes achieve much higher throughputs than conventional electronic packet switching (EPS)-based 5G networks. Although the lack of buffers in OPS switches causes higher packet losses than in EPS, it also leads to very low end-to-end delays due to no queueing delays, helping the congestion control algorithms running at the ends of the path (e.g., inside TCP) to overcome those losses. The performance analysis showcases that OPS achieves lower flow completion times, higher throughputs, and lower probability of missing per-packet end-to-end deadlines. It is also shown that this superior performance of OPS over EPS increases asymptotically in favor of OPS as the traffic loads in the network grow higher.

The paper “Soft-Stacked PON for Soft C-RAN” by Weisheng Hu, Lilin Yi, Hao He, Xuelin Yang, Zhengxuan Li, Meihua Bi, Kuo Zhang, Haiyun Xin, Yuan Liu, and Weijia Du from Shanghai Jiao Tong University, Shanghai University, and Hangzhou Dianzi University considers the application of cloud-based processing for 5G systems. In particular, the aspect of modifying the functional split in the fronthaul architecture, termed “soft C-RAN,” is studied. Two approaches are developed: one using AWG routers together with tunable lasers and the other using wavelength selective switches. Their transmission performance with cost-effective directly modulated lasers (DMLs) is shown, using a delay interferometer (DI) to manage the DMLs’ chirp and fiber dispersion. The result is significant for the future soft C-RAN and stacked PON.

The article entitled “Cost-Minimized Design for TWDM-PON-Based 5G Mobile Backhaul Networks” by Hao Chen, Yongcheng Li, Sanjay K. Bose, Weidong Shao, Lian Xiang, Yiran Ma, and Gangxiang Shen of Soochow University, the Indian Institute of Technology Guwahati, and China Telcom Co. Ltd. considers the deployment problem of many small cells that require very high capacity, low cost, and reliable backhaul connections. The optimized design of a backhaul network for a 5G mobile system based on TWDM-PON is presented. For this, both equipment and deployment costs are considered, and the design satisfies various network constraints. A K-means clustering based algorithm is proposed for the optimal solution. The strategies of using multi-stage remote nodes (RNs) and cable conduit sharing are applied to further reduce the cost. The results suggest that the proposed methods provide substantial improvement over traditional methods.

The contribution by Xinbo Wang, Lin Wang, Cicek Cavdar, Massimo Tornatore, Gustavo B. Figueiredo, Hwan Seok Chung, Han Hyub Lee, Soomyung Park, and Biswanath Mukherjee from the University of California Davis, KTH Royal Institute of Technology, the Federal University of Bahia, and ETRI, entitled “Handover Reduction in Virtualized Cloud Radio Access Networks Using TWDM-PON Fronthaul,” discusses a handover optimization problem leveraging a concept of a virtualized base station (V-BS) that can be formed with software-defined radio (SDR) and coordinated multi-point transmission-reception (CoMP) in a virtualized C-RAN. The main idea of V-BS is to support virtualized resource allocations on a per-user basis, consisting of joint radio transmissions to and from multiple radio units (RUs). In this arrangement, a user can be provided with a mobile service over multiple RUs without handovers within the group. The proposed approach can significantly reduce the number of handovers, handover delays, and failure rates, while increasing the throughputs for fronthaul services.

The paper “Fronthaul Based on Pulse-Width Modulation in RSOA WDM PONs With Broadband and Coherent Seeds” by A. Gatto, P. Parolari, L. Combi, U. Spagnolini, R. Brenot, and M. Martinelli of Politecnico di Milano and the III-V Lab reports an efficient fronthaul optical transmission technique using pulse-width modulation (PWM)-based analog optical transmission. In this approach the radio signal is downconverted to an intermediate frequency (IF) to reduce the channel capacity requirement, and the IF radio signal is converted to a PWM format to be transmitted in a PON fronthaul. An uplink transmission is derived from a downlink optical signal using a reflective optical semiconductor amplifier (RSOA). As an experimental result, this PWM fronthaul technique achieves higher than a 30 dB E_s/N₀.

The article by Mu Xu, Jhih-Heng Yan, Junwen Zhang, Feng Lu, Jing Wang, Lin Cheng, Daniel Guidotti, and Gee-Kung Chang of the Georgia Institute of Technology and the Institute of Photonics Technologies, Taiwan, “Bidirectional Fiber-Wireless Access Technology for 5G Mobile Spectral Aggregation and Cell Densification,” suggests an interesting solution to avoid optical beating interference (OBI) when multiple RRHs transmit uplink optical signals using the same wavelength in a PON. Such OBI can form a broad spectrum after photodetection in both optical homodyne and heterodyne detection schemes. The authors point out that heterodyne demodulation can be marginally better to mitigate OBI because a part of an OBI spectrum near the zero frequency in the electrical signal spectrum can be filtered out as the main signal spectrum appears around the IF introduced by the optical heterodyne detection. The paper presents experimental results of $20\times 80$ and $16\times 80\mathrm{MHz}$ baseband transmissions for downlink and uplink, respectively.

The paper “Ultra-broadband Photonics-Based RF Front-End Toward 5G Networks” by A. L. M. Muniz, R. M. Borges, Regivan N. Da Silva, D. F. Noque, and Arismar Cerqueira S., Jr., from the Brazilian National Institute of Telecommunications proposes a novel ultra-broadband photonics-based RF front-end and experimentally verifies its performance analysis up to 38 GHz. A feature of the proposed RF front-end is its optical frequency multiplication based on an external modulation technique and four-wave mixing for enabling RF upconversion and amplification. Broadband operation and tunability from DC to 38 GHz is experimentally verified for three potential frequency bands for 5G networks.

In the paper “Modeling Energy Performance of C-RAN With Optical Transport in 5G Network Scenarios” by Matteo Fiorani, Sibel Tombaz, Jonas Mårtensson, Björn Skubic, Lena Wosinska, and Paolo Monti of KTH Royal Institute of Technology, Ericsson Research, and Acreo Swedish ICT, the topic of energy performance of four mobile network architectures is considered. Each architecture uses a different baseband processing functional split. The analysis includes both Long-Term Evolution (LTE) and 5G radio access technologies and uses optical wavelength division multiplexing (WDM), with both coherent and direct detection transmission. Results show that with LTE radio interfaces the energy consumption of the transport network is a small fraction of the overall network power consumption. As a result fully centralized radio architectures are viable for LTE. On the other hand, 5G C-RAN, if not carefully designed, might have excessive transport energy consumption.

We certainly hope that the articles of this special issue will serve as a useful resource for researchers who would like to get up-to-date views on the latest research efforts. At this point, we would like to thank the JOCN Editors-in-Chief, Pat Iannone and Ori Gerstel, for endorsing this special issue preparation and the JOCN editorial staff (especially Keith Jackson) for their assistance and support.