Cloud radio access network (C-RAN) has become a promising scenario to accommodate high-performance services with ubiquitous user coverage and real-time cloud computing using cloud BBUs. In our previous work, we implemented cross stratum optimization of optical network and application stratums resources that allows to accommodate the services in optical networks. In view of this, this study extends to consider the multiple dimensional resources optimization of radio, optical and BBU processing in 5G age. We propose a novel multi-stratum resources optimization (MSRO) architecture with network functions virtualization for cloud-based radio over optical fiber networks (C-RoFN) using software defined control. A global evaluation scheme (GES) for MSRO in C-RoFN is introduced based on the proposed architecture. The MSRO can enhance the responsiveness to dynamic end-to-end user demands and globally optimize radio frequency, optical and BBU resources effectively to maximize radio coverage. The efficiency and feasibility of the proposed architecture are experimentally demonstrated on OpenFlow-based enhanced SDN testbed. The performance of GES under heavy traffic load scenario is also quantitatively evaluated based on MSRO architecture in terms of resource occupation rate and path provisioning latency, compared with other provisioning scheme.
© 2016 Optical Society of America
In recent decades, the amount of traffic handled by wireless networks will have increased from under 3 exabytes in 2010 to over 190 exabytes by 2018, on pace to exceed 500 exabytes by 2020 . The proliferation of smart mobile devices and high-definition video applications is forcing the wireless networks from second/third generation (2G/3G) to 4G and beyond. The 5G has become a next major phase of mobile telecommunications  from the perspective of operators and vendors. The vision of 5G should be able to implement a future with both user-services and machine-type communications where access to information and sharing of data is available anywhere and anytime to anyone and anything . To adapt the rapid evolution of 5G mobile network, the cloud radio access network (C-RAN) is a paradigm introduced by operators which aims to reduce capital and operating expenditure and enhance real-time cloud computing while offering better services .
Nowadays, the interaction between remote radio head (RRH) and baseband unit (BBU) or resource schedule among BBUs in cloud have become more frequent and complex with the development of system scale and user requirement. It can promote the networking demand among RRHs and BBUs and force to form elastic optical fiber switching and networking  due to the characteristics of high bandwidth, low cost and transparent transmission. The elastic optical network (EON) is implemented by using the orthogonal frequency-division-multiplexing technology , which assigns necessary spectral resources with a fine tailored granularity for different user service requests, and provides highly-available and cost-effective connectivity . In such network, the multiple stratum resources of radio, optical and computational unit have interweaved with each other. The traditional architecture cannot efficiently enough implement the resource optimization and scheduling for the high-level service guarantee due to the communication obstacle among them.
Additionally, with the rapid evolution of high-bandwidth services (e.g., high-definition video on line), many delay-sensitive services need a high-level end-to-end quality of service (QoS) guarantees . The lower latency and higher capacity have become the major challenges for 5G transport networks in terms of level of 100μs and 100Gbps respectively . The proprietary wireless or optical network hardware appliances deployed as the communication infrastructure are difficult to accommodate the new applications with low costs and higher efficiency. Network functions virtualization (NFV)  aims to address these problems by implementing the network function running as software on generic hardware, which can be consolidated onto industry standard elements, such as switches, computing and storage. NFV can provide tools enabling the creation of infrastructure with a higher level of abstraction with the great advantage to significantly increase the flexibility of network. It causes that the original network and application functions can be partitioned into basic elements  in the fronthaul network. For instance, the operator could place a certain number of servers at base station to implement the functions, such as virtual baseband processing, virtual evolved packet core. Moreover, virtual path computation element  can be deployed on demand to keep the quality of the virtual network functions (e.g., in terms of latency, dedicated algorithms, etc.), which is run as a software application on a cloud computing environment (e.g., a virtual machine). In addition, as a unified software orchestration architecture, software defined networking (SDN) [13, 14 ] including OpenFlow protocol makes maximum flexibility and provides a centralized control over different resources for the joint optimization of functions and resources in a global view [15–17 ]. Therefore, it is very important to apply SDN technique with NFV to control and optimize the resource assignment in such environment.
Service optimization in data center optical networking has been studied with SDN control in our previous works [18–20 ]. Reference  researches the data center service localization and provides the data center application resource closer to users through the virtual migration. The works in [19, 20 ] propose the architecture of elastic optical networking among data center, consider the time factor of data center services, and optimize the optical network and application resource globally using time-aware enhanced software defined networking (TeSDN). Those researches have studied the issue of resource optimization in the data center networking scenario. Based on the research achievement, this paper extends to consider the multiple dimensional resources optimization of radio, optical and BBU processing in cloud-based radio over optical fiber networks of 5G age. The cross stratum optimization (CSO) using SDN between optical network and application stratum resources is proposed to satisfy the QoS requirement in the previous work [8, 21 ]. In view of this, this paper proposes a novel multi-stratum resources optimization (MSRO) architecture with network functions virtualization for cloud-based radio over optical fiber networks (C-RoFN) using software defined control. Based on the proposed architecture, multiple stratum resources of radio frequency, optical and BBU can be evaluated and allocated integrally, thus a global evaluation scheme (GES) is first introduced for MSRO in C-RoFN. The MSRO can enhance the responsiveness to dynamic end-to-end user demands and globally optimize radio frequency, optical and BBU resources effectively to maximize radio coverage. The efficiency and feasibility of the proposed architecture are experimentally demonstrated on OpenFlow-based enhanced SDN testbed . The performance of GES under heavy traffic load scenario is also quantitatively evaluated based on MSRO architecture in terms of resource occupation rate and path provisioning latency, compared with other provisioning scheme.
The rest of this paper is organized as follows. Section 2 describes the MSRO architecture. The global evaluation scheme based on the network architecture is introduced in Section 3. Then we set up the testbed and propose the numeric results and analysis in Section 4. Section 5 concludes the whole paper by presenting our contribution and future work on this area.
2. MSRO architecture for software defined C-RoFN
2.1 Motivation and problem statement
The C-RAN aggregates all base stations computational resources into a cloud BBU pool, while the distributed radio frequency signals are collected by RRH and transmitted to the cloud platform through optical transmission . A RRH executes radio frequency functionalities of a base station (e.g., small cell), while the BBU handles base band processing functions. Fronthaul  is defined as the segment in RRH and BBU and to split the radio base stations, geographically distributed to provide radio coverage. As the typical technology, the EON meets the interconnection requirement and thus forming the enhanced C-RAN connected with EON which is called cloud-based radio over optical fiber networks (C-RoFN). Due to the exponentially growing number of mobile internet users, drastically increased mobile applications and more data-rich Internet content, the interaction between RRH and BBU or resource schedule among BBUs in cloud have become more complex, thus the quality of services cannot be guaranteed in traditional structure. In addition, the SDN using OpenFlow protocol has been extensively researched in terms of IP networks and optical networks, especially including the optical packet/burst switching, fixed and flexi-grid in the access, metro and backbone networks [11–16 ]. The MSRO architecture with SDN control can solve this issue well. It virtualizes the C-RoFN infrastructure as the various kinds of resource cloud through NFV, and then such resources can be orchestrated through the SDN controllers in a unified way among radio, optical and BBU domains. In traditional architecture, the service provisioning schemes just consider one kind of resource, e.g., radio or BBU. Few schemes can globally schedule the multiple stratum resources under the real architecture and environment. So, the GES is introduced based on the proposed architecture. The motivations for MSRO architecture in software defined C-RoFN can break the limit among radio, optical and BBU domain, implement the multiple layer integration and cross stratum optimization based on OpenFlow-enabled C-RoFN with SDN orchestration, which can allocate and optimize the radio, optical network and BBU resources efficiently in a control manner of open system (with GES).
2.2 MSRO architecture for C-RoFN
The MSRO architecture with network functions virtualization in software defined C-RoFN is shown in Fig. 1 . The EON is arranged to interconnect the cloud BBUs, which deployed network and processing (e.g., computing and storage) stratum resources respectively. The distributed RRHs are interconnected and converged into EON, which allocates the customized spectrum with finer granularity for radio signals. Note that, the C-RoFN contains three stratums: radio resource, optical spectrum resource, and BBU processing resource. The networking mode for multi-stratum resources optimization extends in two directions. One is from the perspective of resource form. Optical and computing resources are interconnected cross optical network and BBU stratums along the east–west direction, which is established as “heterogeneous-cross-stratum”. The other is from the perspective of carrying capacity. The interconnection and networking of multiple layers are established along longitudinal direction, which is called “multi-layer-carried”. Based on the above-mentioned networking mode with network functions virtualization, three MSRO applications in this architecture are formed, i.e., the interaction between RRHs (e.g., collaborative radio), the service from RRH to BBU and resource schedule among BBUs (e.g., virtual resource migration inter-BBU). The logical relationship among the networking modes and application scenarios for C-RoFN is shown in Fig. 1. Every resource stratums can be software defined with OpenFlow protocol (OFP) and controlled by a radio controller (RC), an optical controller (OC) and a BBU controller (BC), respectively. To realize the heterogeneous resources control for MSRO with OFP, OpenFlow-enabled RRH with OFP agent are needed, which are referred to as OF-RRH. Note that we can choose various modulation formats and radio frequencies in OF-RRH through the OFP agent. The significance of MSRO architecture in C-RoFN are threefold. Firstly, the MSRO can emphasize the interaction between the RC and OC to overcome the interworking obstacles deriving from multi-layer overlaid networks and it effectively realizes vertical integration between radio frequency and optical network. Secondly, multiple stratum resources can be merged through OC and BC’s cooperation with horizontal merging, while achieving global cross stratum optimization of optical network and BBU resources. Thirdly, based on the integration in two directions, the global resource assess and assignment can realize under the MSRO architecture to optimize the radio, optical and BBU resources in order to enhance the end-to-end QoS.
2.3 Functional models of MSRO for software defined C-RoFN
To achieve the functional architecture, the RC, OC and BC have to be extended to support MSRO and shown in Fig. 2 . Note that, the OFP agent software embedded in OF-BVOS  maintains optical flow table and models node information as software and maps the content to control physical hardware. For the MSRO control of C-RoFN, flow entry of OFP in flow table is extended. In this architecture, the rule is extended as the in/out port, channel spacing, grid, central frequency, spectrum bandwidth, radio frequency which are the main characteristics of C-RoFN. The action of node mainly includes four types: add, switch and drop a path to the port/label with specified adaption function (e.g., modulation format), and delete a path to restore the original state of equipment.
- • Radio controller. The radio frequency monitoring module in RC obtains and manages virtual radio resource in RRH, while the radio frequency allocation module perform radio frequency assignment for the computed path by using OFP. The information can be interacted between the RC and OC through radio-optical interface.
- • BBU controller. The BC obtains BBU resource information periodically or based on event-based trigger through a BBU monitoring module. To conveniently perform the path computation with the CSO of optical and BBU processing stratum resources, the CSO agent in BC can provide the information of BBU computing resource utilization periodically and receive the results from the OC, while interacting with OC through optical-BBU interface.
- • Optical controller. When the service request is arrival, the MSRO control module performs the global evaluation scheme considering radio, optical and BBU resources (will be discussed in the next section), which can decide which BBU node is the destination for resource migration or accommodate for corresponding RRH, assignment application processing resources and determines the location of application or where to migrate virtual machines. The services arrive at the network following a Poisson process. The interval duration time T of services follows the negative exponential distribution, i.e., . Here, the indicates the average arrival rate of services, while means average interval duration of services. Also, the service duration v follows the negative exponential distribution, i.e., . The means the average service rate, while represents the average service time. Then it provides this request to path computation element (PCE) module in turn, including the request parameters (e.g., latency and bandwidth), and eventually returning a success reply including the information of the provisioned path. After receiving the processing resources information from the BC, the end-to-end path computation from RRH to BBU can be completed in PCE module considering CSO of the optical and BBU resources. When the path is setup successfully, the information of the path is conserved into the database management in the OC, which can interact with network virtualization module and store virtual network and BBU resources for MSRO.
3. Global evaluation scheme
In the C-RoFN, the multiple stratum resources of radio network, optical network and BBU processing resources are deployed in this scenario used for the optimization. The traditional resources evaluation schemes assess the resource utilization only considering one kind of resources partly. In this section, based on functional architecture, we propose a global evaluation scheme in the OC to realize the multiple stratum resources optimization cross radio, optical and BBU stratum to guarantee the QoS demands. The GES includes two phases. In the first phase, GES can choose the best destination BBU according to multiple stratum resources using global evaluation factor. The second phase performs service provisioning with continuous spectrum and radio frequency assignment for the destination. The two phases just correspond two kinds of resource optimization in two directions. In the first phase, the destination node selection will use multiple stratum resources optimization with horizontal merging, while achieving global cross stratum optimization of optical network and BBU resources. In the second phase, after the destination choice, the service provisioning with radio and spectrum assignment should be performed, which provides the multiple layer resource optimization from multi-layer overlaid networks.
3.1 Network modeling
The MSRO architecture in software defined C-RoFN is represented as G (V, V’, L, L’, F, F’, A), where V = and V’ = denote the set of OpenFlow-enabled optical switching and RRH nodes, respectively. In addition, L = and L’ = indicate the set of bi-directional fiber links between nodes in V and V ’. Also, F = and F’ = are the set of optical spectrum and radio frequency on each fiber link and A denotes the set of BBU nodes, while V, V’, L, L’, F, F’ and A represent the number of network nodes, links, the spectrum and radio slot and BBU nodes respectively. In each BBU server, two time-varying BBU processing stratum parameters describe the service condition of computing and storage resources, which are comprised of storage utilization modeled storage and CPU usage. In addition, the parameters in optical network consist of the hop Hp of each candidate path, and the occupied network bandwidth weight Wl of each link, which are related to traffic load of the corresponding link. The radio parameters contain the symbol rate Br and radio frequency Fr of current radio signal. The BBU can provide the required computing and storage resources to enhance the experience of QoS in the fronthaul area. Therefore, the service request contains of the source node of BBU and the needed network and BBU application resources for service accommodation. For each request from source node s, it can be translated into the needed network and processing resources. Note that these resources include the required network bandwidth b and CPU and storage processing resources for simplicity. We denote the ith service request described above as . In addition, according to the service request and status of resources, the appropriate BBU server can be chosen as the destination node based on the scheme.
3.2 Phase I: destination selection with global evaluation factor
We propose a global evaluation scheme based on the functional architecture of MSRO described above in the optical controller.
The first phase of GES is to select the service destination of BBU and evaluate the network status coordinated processing resources. Before a new service request arrives, the service demand has included service parameters, i.e., . The GES can select the BBU based on the processing status collected from BBU, and the radio and optical condition provided by the RC and OC respectively. Note that, the multiple parameters from various stratums are hard to be evaluated because of different dimensions. To measure the choice rationality of service provisioning, we define as the global evaluation factor which considers all multi-stratum parameters. For radio, optical network and BBU occupation, there are several parameters to influence the system performance. Note that, the BBU parameters include CPU usage, storage utilization, processing scheduling and so on in BBU stratum. For simplicity, two processing parameters, CPU usage and storage utilization describe the current usage of BBU resource, which can be easily obtained through the open interfaces. In addition, optical network parameters are comprised of the traffic engineering weight Wl of the current link and the hop Hp of the candidate path. The traffic engineer weight means the rate of occupied bandwidth and the total network bandwidth in the link. It indicates that smaller value of traffic engineer weight leads to the better capacity of carrying new services. Note that, we assume the total network bandwidths are the same among the links, which is convenient for the networking of optical nodes. So, if the values of traffic engineer weight are the same, the occupied bandwidths are also the same. The definition of traffic engineer weight is used for the measure of traffic load balancing in optical network. Aimed at this issue, the vacant link with smaller value of traffic engineer weight should be selected to enhance the degree of load balancing. The radio parameters contain the symbol rate Br and radio frequency Fr of current radio signal. Therefore, the overall BBU function fac with the BBU stratum parameters of current each server is expressed as dimensionless Eq. (1), where is the adjustable proportion between storage and CPU usage, and these parameters are normalized to meet the linear relationship between them. In BBU function, the CPU and storage utilization are hard to be evaluated because of different dimensions. To measure the choice rationality of BBU processing, the adjustable proportion weight is used to adjust the proportion of them.
In addition, optical network function fbc with parameters of current each optical node is expressed as dimensionless Eq. (2), while radio function fcc is Eq. (3). The radio function is used as a cost weight to measure the carrying capacity of radio link. In Eq. (3), the Br indicates the symbol rate while the Fr means the radio frequency of current radio signal. Bigger value of symbol rate (i.e., Br) leads to bigger value of carrying capacity, while bigger value of radio frequency (i.e., Fr) leads to smaller value of carrying capacity of radio link for services. Therefore, the radio function is shown as Eq. (3). The K candidate BBU server nodes with the first K minimum of processing functions are chosen and expressed as the set Fa = . Here, the a1, a2,…, ak means the K candidate BBU server nodes with the first K minimum of processing functions. Then the candidate lightpaths between source of optical network and each candidate BBU server can be calculated with minimum optical network function and denoted as Fb = . Similarly, the b1, b2,…, bk means the candidate lightpaths between source of optical network and each candidate BBU server can be calculated with minimum optical network function. The candidate radio frequency links between source and optical network node with minimum radio function are indicated as Fc = associated with the K candidate paths. So the global evaluation factormeets the Eq. (4) as follows, where and γ are the adjustable weights among the BBU, optical and radio parameters. The maximum of set Fa, Fb, Fc as the denominator can insure each value of the equation part between zero and one. Note that, the traffic engineering weight Wl can be obtained in the current optical network, which is the rate of occupied bandwidth and the total network bandwidth. While the and γ, as the preset proportion, are the adjustable weights among the BBU, optical and radio parameters. The value of them can be set up ahead of time to describe the importance of these parameters. They are constant in the equation computation, since they have been set before the calculation already.
In the first phase, according to BBU resource utilization, the GES first chooses the best K candidate BBU nodes in BBU stratum for radio signals and continuous spectrum path. Then in radio and optical stratums, the node with minimumvalue based on the global evaluation factor will be selected from the K candidates as the optimal destination node.
3.3 Phase II: radio and spectrum allocation
After the destination selection, the second phase of GES is service provisioning with radio and spectrum allocation. We assume that the BBU node includes computing and storage resources, while the BBU pool can be seen as a data center. Meanwhile, the network devices support network function virtualization, where network function runs as software on generic hardware, which can be consolidated onto industry standard elements, e.g., switches, computing and storage.
In radio and spectrum allocation phase, three dimensional resources are considered in this phase, which contain radio frequency, spectrum and link. We accommodate the service considering radio frequency assignment with available spectrum first, and then use other fit spectrum resource. The allocation process is illustrated with an example shown in Fig. 3 , which considers a simple network with six nodes shown in Fig. 3(a). The available radio frequency on each fiber link is divided into 9 radio frequency slots (FS) modulated with each optical spectrum, while the labels of FSs are arranged in increasing order from 1 to 9. On the spectrum dimensionality, we only consider optical spectral resource of spectrum label 1 and 2 for simplicity. In the topology, we assume that service requests arrive in time order as shown with different colors. In initial state, the first three demands SR1, SR2 and SR3 choose the route la, b, lb, e and la, f, lf, e, ld, e and la, b, lb, c, lc, e as the path respectively and each required FS resources (i.e., 3, 6, 5) are reserved on each path modulated with spectrum 1 because all resources are not occupied in the initial context. When SR4 arrives at node A, it selects the route la, b, lb, c, lc, d first with spectrum 1 and has to be blocked because 4 consecutive FSs cannot be available along this path with spectrum 1. Then SR4 chooses the spectrum 2 as the modulated optical spectrum and first four FSs will be utilized for allocation. Then the path can be established with spectrum and modulated radio frequency allocation through OpenFlow protocol between the source and destination nodes after the choice of the BBU.
4. Experimental demonstration and performance evaluation
To evaluate the efficiency of the proposed architecture, we set up an EON with software defined C-RoFN based on our testbed, as shown in Fig. 4 . In data plane, two analog RoF intensity modulators and detect modules are utilized, which driven by a microwave source working at 40GHz frequency to generate double sideband. Four OpenFlow-enabled elastic ROADM nodes are equipped with Finisar BV-WSSs in the EON. We use Open vSwitch as software OFP agent according to the API to control the hardware and interact between controller and radio and optical nodes. In addition, OFP agents are used to emulate other nodes in data plane to support the MSRO with OFP. The BBU pool and OFP agents are implemented on many virtual machines created by VMware ESXi V5.1 running on IBM X3650 servers. The virtual operation system technology makes it easy to set up experiment topology for large scale extension. For OpenFlow-based MSRO control plane, the OC server is allocated to support the proposed architecture and deployed by means of three virtual machines for MSRO control, network virtualization and PCE strategy as plug in, while the RC server is used as radio frequency resource monitor and assignment. The BC server is deployed as CSO agent to monitor the computing resources from BBUs. Each controller server controls the corresponding resources, while the database servers are responsible for maintaining traffic engineering database, connection status and the configuration of the database. We deploy the service information generator related with the controllers, which implements batch C-RoFN services for experiments.
Based on the testbed, we have designed and verified experimentally MSRO for service in C-RoFN. The experimental results are shown in Fig. 5 and Fig. 6 . Figures 5(a)-5(b) present the signaling procedure for MSRO using OFP through a Wireshark capture deployed in OC and RC respectively. Note that the new messages types for C-RoFN will be studied and defined to support new functionalities in our future work. As shown in Figs. 5(a)-5(b), 10.108.67.21, 10.108.50.74 and 10.108.49.14 denote the IP addresses of the RC, OC and BC respectively, while 10.108.49.23 and 10.108.49.24 represent the IP addresses of related OF-BVOSs respectively. The features request message is responsible for monitoring by regularly querying OF-BVOSs about the current status. The OC obtains the information from OF-BVOSs via features reply. When the service request arrives, the RC sends the request for MSRO via UDP message, where we use UDP message to simplify the procedure and reduce the performance pressure of controllers. After receiving the resources information from the interworking, the OC performs the GES to compute the paths considering multiple stratum optimization of radio, optical network and BBU resources, and then reserve the optimal radio frequency, spectrum and processing resources for the service provisioning. After completing GES, the OC and RC provision spectral path and assign the radio frequency to control the corresponding nodes via flow mod message. Receiving the setup success reply via packet in, the RC responds the MSRO success reply to BC and updates the computational usage to keep the synchronization.
In the established testbed, we consider the real application scenarios and the complexity of the experiment setup, and establish the service demand realized in service generator for simplicity. For resource visualization, the front-end interface of the testbed is shown partly in Fig. 6, which can be used to demonstrate the service provisioning according to the accommodation status of radio, spectrum and BBU resources. As shown in Fig. 6, two service requests from RRH to BBU can be provisioned in the MSRO architecture concurrently. We can see that two services are accommodating on the various paths, which can be presented in the virtual topology of interface. It can be seen clearly that the bottom of Fig. 6 indicates the current network bandwidth and processing resource status of BBU servers, and corresponding routing information of serving paths, which include detailed path, service bandwidth and related service order. The spectrum of lightpath for analog C-RoFN is reflected on the filter profile, as shown in Fig. 7(a) . The small figure in lower right corner of Fig. 7(a) means the shrinking version of the main figure. The radio signals can be modulated on the spectral channel with MSRO.
We also estimate the performances of MSRO with GES under heavy traffic load scenario and compare it with the traditional CSO scheme  in terms of resource occupation rate and path provisioning latency using virtual machines. The heavy traffic load scenario means the network under the batch service requests, where the traffic load is from 40 Erlang to 150 Erlang. The traditional CSO scheme accommodates the services request only considering CSO of optical network and application stratum resources, and transports it from source to destination BBU server node. Here, the classical first fit strategy is used for the spectrum allocation. The requests are setup with bandwidth randomly distributed from 500MHz to 40GHz, where the spectrum slots in elastic optical network is 6.25GHz. We assume the CPU usage in BBU is selected randomly from 0.1% to 1% for each service request, while the storage resource in server is occupied from 1GB to 10GB for each demand. The services arrive at the network following a Poisson process and results have been extracted through the generation of 100,000 demands per execution. Each service demand’s duration and inter-arrival interval duration follow the negative exponential distribution. To calculate the Eqs. (1)-(4) , there are several preset weights in the formulas. The is the preset weight between CPU and storage to measure the importance of them. The and γ are the preset weights among BBU, optical and radio parameters. For the proposed GES, we preset the values of adjustable weight , and γas 50%, 33% and 33% respectively to avoid the experiment complexity in the simulation settings. The time-varying parameters in the scheme will be researched in the future work. According to CPU usage and storage utilization , GES first chooses the best K candidate BBU nodes in BBU stratum by calculating with Eq. (1). In radio and optical stratum, the radio and optical network utilization with the candidate nodes should be computed using Eqs. (2)-(3) , while the corresponding value based on the global evaluation factor can be computed using Eq. (4). Then the node with minimum value will be selected from the K candidates as the destination node for the services according to radio, optical and BBU utilization. The first fit strategy can be performed as the radio frequency and spectrum assignment for the provisioning path between the source and destination after the node choice. Figures 7(b)-7(c) compare performances of two schemes in terms of resource occupation rate and path provisioning latency. Resource occupation rate reflects the percentage of occupied resources to the entire radio, optical network and BBU resources. As shown in the figure, the GES can enhance the resource occupation rate effectively than the other scheme, especially when the network is heavily loaded. The reason is that the GES can globally optimize the radio, optical and BBU stratum resources to maximize radio coverage. Figure 7(c) shows the GES reduces the path provisioning latency compared to the other. That is because the GES chooses the destination BBU before the service arrives, which leads to low computation and provisioning time being consumed.
To satisfy the service accommodation in 5G age, this paper proposes a MSRO architecture with network functions virtualization for software defined C-RoFN. Meanwhile, the GES is introduced for MSRO based on the proposed architecture, which can evaluate optimal destination BBU server and perform the radio and spectrum allocation for the services. The functional architecture is described in this paper. The efficiency of MSRO is demonstrated on our OpenFlow-based enhanced SDN testbed. We also quantitatively evaluate the performance of GES under heavy traffic load scenario in terms of resource utilization rate and provisioning latency, and compare with the traditional CSO scheme. The results indicate the MSRO with GES utilizes multiple stratum resources of radio frequency, optical network and BBU processing effectively and enhance end-to-end responsiveness of user services.
Our future work includes two aspects. One is to improve the GES performance with the time-varying parameters and consider the scalability issues of C-RoFN with a large scale network topology with multi-domain. The other is to develop the new messages types to support new functionalities for MSRO in C-RoFN on OpenFlow-based testbed.
This work has been supported in part by National Natural Science Foundation of China (NSFC) project (61501049), Ministry of Education-China Mobile Research Foundation (MCM20130132), the Fundamental Research Funds for the Central Universities (2015RC15), Fund of State Key Laboratory of Information Photonics and Optical Communications (BUPT), China (IPOC2015ZT01), and funded by State Key Laboratory of Advanced Optical Communication Systems Networks, China.
References and links
1. J. G. Andrews, S. Buzzi, W. Choi, S. V. Hanly, A. Lozano, A. C. K. Soong, and J. C. Zhang, “What will 5G be?” IEEE J. Sel. Areas Commun. 32(6), 1065–1082 (2014). [CrossRef]
2. A. Oliva, X. Perez, A. Azcorra, A. Giglio, F. Cavaliere, D. Tiegelbekkers, J. Lessmann, T. Haustein, A. Mourad, and P. Iovanna, “Xhaul: toward an integrated fronthaul/backhaul architecture in 5G networks,” IEEE Wirel. Commun. 22(5), 32–40 (2015). [CrossRef]
3. M. Fiorani, P. Monti, B. Skubic, J. Martensson, L. Valcarenghi, P. Castoldi, and L. Wosinska, “Challenges for 5G transport networks,” in Proceedings of IEEE International Conference on Advanced Networks and Telecommuncations Systems (ANTS 2014), (IEEE, 2014), 1–6.
4. J. Wu, Z. Zhang, Y. Hong, and Y. Wen, “Cloud radio access network (C-RAN): A primer,” IEEE Netw. 29(1), 35–41 (2015). [CrossRef]
5. K. Tanaka and A. Agata, “Next generation optical access networks for CRAN,” in Proceedings of Optical Fiber Communication Conference (OFC 2015), (Optical Society of America, 2015), paper Tu2E.1.
6. M. Jinno, H. Takara, B. Kozicki, Y. Tsukishima, Y. Sone, and S. Matsuoka, “Spectrum-efficient and scalable elastic optical path network: architecture, benefits, and enabling technologies,” IEEE Commun. Mag. 47(11), 66–73 (2009). [CrossRef]
7. O. Gerstel, M. Jinno, A. Lord, and S. J. B. Yoo, “Elastic optical networking: a new dawn for the optical layer?” IEEE Commun. Mag. 50(2), s12–s20 (2012). [CrossRef]
8. H. Yang, J. Zhang, Y. Zhao, Y. Ji, J. Han, Y. Lin, and Y. Lee, “CSO: cross stratum optimization for optical as a service,” IEEE Commun. Mag. 53(8), 130–139 (2015). [CrossRef]
9. A. Pizzinat, P. Chanclou, F. Saliou, and T. Diallo, “Things you should know about fronthaul,” J. Lightwave Technol. 33(5), 1077–1083 (2015). [CrossRef]
10. R. Nejabati, S. Peng, M. Channegowda, B. Guo, and D. Simeonidou, “SDN and NFV Convergence a Technology Enabler for Abstracting and Virtualising Hardware and Control of Optical Networks (Invited),” in Proceedings of Optical Fiber Communication Conference (OFC 2015), (Optical Society of America, 2015), paper W4J.6.
11. H. Yang, J. Zhang, Y. Ji, R. Tian, J. Han, and Y. Lee, “Performance evaluation of multi-stratum resources integration based on network function virtualization in software defined elastic data center optical interconnect,” Opt. Express 23(24), 31192–31205 (2015). [CrossRef] [PubMed]
12. R. Vilalta, R. Muñoz, R. Casellas, R. Martínez, V. López, and D. López, “Transport PCE Network Function Virtualization,” in Proceedings of European Conference and Exhibition on Optical Communications (ECOC 2014), (Optical Society of America, 2014), paper We.3.2.2.
13. L. Liu, W. R. Peng, R. Casellas, T. Tsuritani, I. Morita, R. Martínez, R. Muñoz, and S. J. B. Yoo, “Design and performance evaluation of an OpenFlow-based control plane for software-defined elastic optical networks with direct-detection optical OFDM (DDO-OFDM) transmission,” Opt. Express 22(1), 30–40 (2014). [CrossRef] [PubMed]
14. M. Channegowda, R. Nejabati, M. Rashidi Fard, S. Peng, N. Amaya, G. Zervas, D. Simeonidou, R. Vilalta, R. Casellas, R. Martínez, R. Muñoz, L. Liu, T. Tsuritani, I. Morita, A. Autenrieth, J. P. Elbers, P. Kostecki, and P. Kaczmarek, “Experimental demonstration of an OpenFlow based software-defined optical network employing packet, fixed and flexible DWDM grid technologies on an international multi-domain testbed,” Opt. Express 21(5), 5487–5498 (2013). [CrossRef] [PubMed]
15. H. Yang, J. Zhang, Y. Ji, Y. Tan, Y. Lin, J. Han, and Y. Lee, “Performance evaluation of data center service localization based on virtual resource migration in software defined elastic optical network,” Opt. Express 23(18), 23059–23071 (2015). [CrossRef] [PubMed]
16. R. Martínez, R. Casellas, R. Vilalta, and R. Muñoz, “Experimental assessment of GMPLS/PCE-controlled multi-flow optical transponders in flexgrid networks,” in Proceedings of Optical Fiber Communication Conference (OFC 2015), (Optical Society of America, 2015), paper Tu2B.4. [CrossRef]
17. F. Paolucci, F. Cugini, N. Hussain, F. Fresi, and L. Poti, “OpenFlow-based flexible optical networks with enhanced monitoring functionalities,” in Proceedings of European Conference and Exhibition on Optical Communications (ECOC 2012), (Optical Society of America, 2012), paper Tu.1.D.5. [CrossRef]
18. H. Yang, Y. Zhao, J. Zhang, Y. Tan, Y. Ji, J. Han, Y. Lin, and Y. Lee, “Data center service localization based on virtual resource migration in software defined elastic optical network,” in Proceedings of Optical Fiber Communication Conference (OFC 2015), (Optical Society of America, 2015), paper Th4G.4. [CrossRef]
19. H. Yang, J. Zhang, Y. Zhao, Y. Ji, H. Li, Y. Lin, G. Li, J. Han, Y. Lee, and T. Ma, “Performance evaluation of time-aware enhanced software defined networking (TeSDN) for elastic data center optical interconnection,” Opt. Express 22(15), 17630–17643 (2014). [CrossRef] [PubMed]
20. J. Zhang, H. Yang, Y. Zhao, Y. Ji, H. Li, Y. Lin, G. Li, J. Han, Y. Lee, and T. Ma, “Experimental demonstration of elastic optical networks based on enhanced software defined networking (eSDN) for data center application,” Opt. Express 21(22), 26990–27002 (2013). [CrossRef] [PubMed]
21. H. Yang, J. Zhang, Y. Zhao, Y. Ji, J. Wu, Y. Lin, J. Han, and Y. Lee, “Performance evaluation of multi-stratum resources integrated resilience for software defined inter-data center interconnect,” Opt. Express 23(10), 13384–13398 (2015). [CrossRef] [PubMed]
22. X. Rao and V. Lau, “Distributed fronthaul compression and joint signal recovery in Cloud-RAN,” IEEE Trans. Signal Process. 63(4), 1056–1065 (2015). [CrossRef]
23. B. Skubic, G. Bottari, A. Rostami, F. Cavaliere, and P. Ohlen, “Rethinking optical transport to pave the way for 5G and the networked society,” J. Lightwave Technol. 33(5), 1084–1091 (2015). [CrossRef]