CLOUD RAN INTRODUCTION In today’s world, the usage of wireless data has been growing and the level of bandwidth content increases, mobile operators exponentially making it necessary for mobile operators to scale are finding it dicult to convert the increased demand for the wireless networks. This is a result of proliferation of new broadband into an increased profit. On the other hand, apps, devices, BYOD, usage of wireless broadband for home introduction of new technologies, expansion of existing network surveillance, public safety etc. Meeting this unabated growth in put Capex and Opex pressure on operators. In our opinion, demand requires cellular operators to scale their wireless virtualization helps operators in overcoming these opposite networks and use of advanced wireless technologies. As mobility MOTIVATIONS AND BENEFITS OF CLOUD RAN Mobile operators also face the challenge for supporting base stations in turn causes interference to each other and thus networks with a spurt of increase or decrease of trac. For e.g., slows the network. The network develops lot of cell edges that if there is some rugby or football event, then the need for act as speed bumps for broadband trac. Apart from the cell bandwidth increases. So the area near the stadium needs to site and hardware cost, C-RAN architecture can provide a support much higher capacity for a single day. Another use case virtually edge free network by enablement of spectral eciency is that in order to optimize the capex, operators are looking for improvement features. option to dynamically use the compute resources such they can be used in business districts during day (where there is high C-RAN architecture increases the spectral eciency of the demand during day) and same resources used in residential area access network. The cell edge performance can be improved by in night. Cloud based deployments of access network helps a large scale deployment of technologies like eICICI and CoMP. operator meet these challenges. These technologies require a low latency communication among participating cells. Decision on scheduling needs to be made in a Cloud RAN (C-RAN) architecture reduces the antenna site very short interval so as to meet LTE air interface requirement as hardware including computing resources without impacting Air well as coordination need across cells. High bandwidth commu- or Core network interface. This is achieved by moving most of nication among centrally hosted participating cells is a key the base station processing to a central location or cloud, element required for the realization of such spectral eciency handling multiple cell sites at the same time. C-RAN architecture methods. can support virtualization of Access Network across multiple Radio Access Technologies. Hosting multiple RAT in virtualized As the computing moves to cloud, resources can be dynamically environment not only assist in launching new RAT but also allocated / de-allocated as per the network trac demand. In upgrading or extending the installed base. This leads to a signifi- traditional networks, operator need to provision the proprietary cant reduction in leasing, maintenance, energy and back up cost hardware at a cell site according to peak trac in each and every for cell sites. cell even though the subscriber base moves within the network based on time of day. Movement of subscribers, specialized With increase in mobile trac, operators need to deploy dense applications, events, IoT linked devices put variable bandwidth and sometimes more dense network. This increase in number of demand on the network. Cloud RAN 1 Average load on a cell is much lower in comparison to peak load hardware goes soft on Capex. Solutions for operators are no in a cell. Load on the entire network is close to average but longer tied to specialized hardware boxes. The same hardware dierent cells experience peak load at dierent points in time. can be used to host dierent technologies or mix of technolo- Virtualization aided elasticity in resource utilization brings gies based on the field requirements. The solution becomes multi-fold benefits. It could mean more bandwidth in a specific more software centric. With specialized hardware boxes out of area, more device support for networks like IoT, scalable access the way, operators get more choices for the software. It not only and core networks, or even could be a combination. Major speeds up the R&D innovation, it reduces the overall cost to the features need not wait for a major upgrade - a key decision solution. factor in traditional networks. Highly competitive market demands agile mode for introductions of new features. Virtual- In our opinion, software centric solution, provided through ized networks aid in adding new features or roll them back Radio Network virtualization, aids in dynamic scaling, faster based on need. introduction of features and services and reduction of cell site equipment and hence saves Capex and Opex for operators. As Virtualization aids in deploying the access network solution for per our analysis, this will lead to many new vendors providing multiple operators as resources can be easily shared. Operator solutions and operators can thus roll out new services with very can install services for specific geography as per need and even less lead time. roll it back. The resources can be scaled and shared by multiple operators. Aricent is working on dierent solutions in C-RAN space. Aricent Cloud RAN oering is a mix of cloud RAN enabling The base station also uses proprietary hardware slowing the software and product engineering services. This oering can be R&D cycle. Cloud architecture is supported on standard IT leveraged to develop and deploy next-generation virtualized servers, switches and storage. Use of standard pieces of wireless network. DIFFERENT C-RAN ARCHITECTURES FOR DIFFERENT DEPLOYMENT NEEDS C-RAN is an architecture choice and allows Node B or Base LTE eNodeB is a layer 3 OSI device. Virtualization of dierent station to be developed in multiple ways. Dierent architectures layers oer dierent solution as shown in the Figure below. cater to dierent deployment needs and have dierent implementation complexity. C-RAN : Architecture Options Virtualized Space Non Virtualized Space Layer 2 Layer 3 S1-U (RLC/MAC, Layer 1 Layer 3 C-RAN (RRC, QAM) PDCP/GTP RRM) Layer 2 Layer 3 Layer 2 C-RAN (RLC/MAC, Layer 1 (RRC, QAM) PDCP/GTP, RRM) Layer 2 Layer 3 (RLC/MAC, Layer 1 C-RAN Layer 1 (RRC, QAM) PDCP/GTP, RRM) Cloud RAN 2 Layer 3 Cloud RAN It is possible to virtualize eNodeB Layer 3 consisting of Radio S1-U hosted on cloud architecture enables Dual cell Resource Control, OAM, X2 interface. PDCP/GTP layer are connectivity whereby mobile can be reached from two cells. considered as Layer 2 in LTE eNodeB architecture. These layers The architecture also enables LWA (LTE – Wi Fi Aggregation) are non-real time in nature and can also be virtualized. whereby Wi-Fi hot spots can be integrated with LTE Cloud RAN solution increasing the deployment capacity by using The architecture allows an option whereby multiple cells in a unlicensed spectrum. dense deployment can be aggregated and provided a single S1 view towards core network. Such architecture option can be In the architecture Layer 3 and S1-U layers are hosted on used in enterprise or dense deployment cases whereby there is centralized IT servers. The servers are connected with antenna a need to improve Core network connection behavior in dense site over Ethernet. Antenna sites have LTE Layer 2 (RLC/MAC, deployed case as well as control the visibility of handovers in Scheduler) and Layer 1 implemented. core network. Cloud RAN Unit RRC, RRM LTE L3 OAM, Single S1/X2 eNB C-RAN PGDCP/GTP VM1 VM2 VM3 VM4 QEMU GTP PDCP PDCP L3 Distribution Call 1 Call 2 Guest Guest Guest Guest Linux Linux Linux Linux KVM VIM Host RT Linux RLC/MAC LTE Phy LTE AP LTE AP LTE AP Layer 3 C-RAN Layer 2 C-RAN In Layer 2 C-RAN architecture, LTE layer 2 and above software using Ethernet. The architectures can be leveraged across is hosted at a central location on standard IT servers. These various deployment scenarios. Layer 2 C-RAN can be further servers connect with the antenna site through IP links such as implemented in following dierent ways. MAC – Split Architecture In the LTE MAC Split architecture, the one-way latency between All the benefits of Layer 3 C-RAN, viz S1/X2 Aggregation, Dual cloud and antenna site is more than a millisecond. In order to Cell Connectivity, LWA Enablement, are also present in this meet LTE Air interface requirement, HARQ portion of MAC is architecture. In addition, the architecture option further moves kept at antenna site whereas MAC scheduler and Layer 3 processing needs at antenna site to centralized IT servers. resides at central location. The architecture is well-suited for wide-area continuous deployment. The distance between cloud and cell site can be further extended using CPRI/OBSAI links from antenna sites. Cloud RAN 3 Cloud RAN Unit RRC, RRM LTE L3 OAM, eNB1 eNB2 C-RAN PGDCP/GTP VM1 VM2 VM3 VM4 QEMU LTE L2 L2 L3 L2 L3 { RLC MAC (Upper) Guest Guest Guest Guest Linux Linux Linux Linux KVM VIM Host RT Linux MAC (Lower) LTE Phy LTE AP Layer 1 Layer 1 Layer 2 MAC Split C-RAN L2+ IN CLOUD ARCHITECTURE In case of the L2+ in Cloud Architecture solution, complete L2+ All the benefits of Layer 3 C-RAN, viz S1/X2 Aggregation, Dual resides in cloud. The architecture necessitate that the one way Cell Connectivity, LWA Enablement, are also present in this latency between antenna and central site is of the order of 100 architecture.