Cloud RAN (Tentative de traduction) Réseau d accès radio basé sur le Cloud Loutfi Nuaymi, Télécom Bretagne / Network, Security and Multimedia Department IRISA/ REOP team Supéléc, Cesson-Sévigné 17 Oct 2013 Plan 1- Context and objectives 2- Definitions and concepts 2.1- full and partial centralization 3- Advantages and benefits of Cloud RAN 3.1- Associate with software-defined networks (SDN) 3.2- Associate with Active Antennas and beamforming 4- Normalization and commercialization 5- Identified research issues (not exhaustive) 6- State of the Art of research works 7- Our Work References page 1
1- Context: the traffic-revenue gap Well known wireless networks Energy Constraints: energy contribution of ICT is XX % of the global energy consumption (Gartner, 2007) Mobile Data traffic increase : x2 every year or two years according to different sources x1000 in around 15 years (but who will verify in 15 years?) Trafic gap : the ARPU increases slowly (or not) * Example of reference: Cisco traffic forecast update, Cisco visual networking index: global mobile data, 2011. Source Fig.: Rit12, Projet Greentouch page 2 1- Context: need for flexibility The density of base stations is very high in urban areas. It is very difficult to add new BS sites in this kind of area Yet, during large periods of time, many Base Station are underused. Flexibility is needed in order to provide very high data rates at a smaller cost. Figure Source : L. Correia, et al, Challenges and Enabling Technologies for Energy Aware Mobile Radio Networks, IEEE Com. Mag. Nov 2010 page 3
1- Context: the cloud paradigm From Zander, Mahonen, IEEE Com. Mag., March 13 : «- History has shown a constant shift between computational paradigms that has adapted to computing and computational bottleneck. - We have gone from the mainframe era in the 1960s, when both communication and processing were expensive, to the PC era in the 1980s and 1990s when processing became cheap, but communication still was a bottleneck. - Now we are moving toward the cloud paradigm, where communication is virtually free, and remote computing can be the most effective solution. page 4 2- Definitions and concepts From [5]: the traditional base station solution uses antennas connected via thick (and lossy) RF cables to the cabinet containing the RF modules and baseband processing. The more recent remote radio head (RRH) or antenna-integrated radio designs place the RF module next to the antenna to reduce cable losses. page 5
2- Definitions and concepts: BBU and RRH The Cloud RAN base stations architecture breaks down the classical base station into: - a BaseBand Unit (BBU), a digital unit that implements the MAC PHY and AAS (Antenna Array System) functionality - a Remote Radio Head (RRH), obtains the digital (optical) signals, converts digital signals to analog, amplifies the power, and sends the actual transmission. Source Fig: http://frankrayal.com/ BS = (local, indoor) BBU + (distant, close to user, outdoor) RRH By making the RRH an active unit capable of converting from analog to digital, operators can now place numerous BBUs in a single geographical point while distributing the Remote Radio Units (RRUs) according to the Radio Frequency (RF) plans. The RRH becomes an intelligent antenna array which not only submits RF signals but also handles the conversion between digital and modular data. page 6 2- Definitions and concepts: BBU and RRH With the BBU/RRH separation, we have a Distributed Antenna System (DAS) Source Fig: R. Qingyang, Cooperative, Green and Mobile Heterogeneous Wireless Networks, ITC23, Sept 2011 With this approach, the radio can be hidden behind active macro antennas on rooftops or tucked into a small space in any building. page 7
2- Definitions and concepts: What is Cloud RAN? Replacing a classical Base Station (GSM Style), with about 100 users, a BBU Pool or Virtual BS Pool (sometimes called Cluster) provides much more capacity and greater coverage (order of magnitude: 10 kms instead of 2kms), Collaborative Radio: Multi-cell Joint scheduling and processing, Well-suited for femtocell extensions. Source Fig: C. Chen, C-RAN: the Road Towards Green Radio Access Network, China Mobile Slides, August 2012 page 8 2- Definitions and concepts: What is Cloud RAN? The overall LTE Cloud Overall Architecture can then be the following: Source Fig: B. Haberland Slides, Base Stations in the Cloud, Sept 2012 MSS-BBU: Multi-site/standard BBU DCC: Decentralized Cloud Controler ex2: enhanced X2 interface page 9
2- Definitions and concepts: C-RAN starting from a leagcy BS Different scenarios are possible Overlay scenario: introducing a new frequency band or a new radio standard to provide larger network capacity. The methods mentioned can be combined together (see figure) Source [4] NGMN, Jan 13 page 10 2.1- full and partial centralization The BBU and RRH separation has two possibilities. (From [1], modified) according to the different function splitting between BBU and RRH, there are two kinds of C-RAN solutions: - Solution 1: one is called full centralization, where baseband (i.e. layer 1) and the layer 2, layer 3 BTS functions are located in BBU; - the other (Solution 2) is called partial centralization, where the RRH integrates not only the radio function but also the baseband function, while all other higher layer functions are still located in BBU. The BBU doesn t include the baseband function but is still called BBU for simplicity. Source Fig: [1] page 11
2.1- Full and partial (or distributed baseband processing) centralization The architecture options are illustrated in this figure. Source Fig: [6] page 12 2.1- full and partial centralization Other Illustration (Chen, CMRI, White Paper, Cite): C-RAN Architecture 2: Partial Centralized Solution Source Fig: [6] cite, see above page 13
2.1- full and partial centralization (Source: Chen CRI White Pap [1]) Advantages of fully centralized C-RAN architecture: - easy upgrading and network capacity expansion, - better capability for supporting multistandard operation, - maximum resource sharing, - more convenient towards support of multi-cell collaborative signal processing. Its major disadvantage is the high bandwidth requirement between the BBU and to carry the baseband I/Q signal between BBU and RRH. (according to CMRI, [1]) In the extreme case, a TD-LTE 8 antenna with 20MHz bandwidth will need a 10 Gpbs transmission rate. page 14 3- Advantages and benefits of Cloud RAN The advantages of Cloud RAN are (1/2): From [1]: - Money-saving: Lower CAPEX (-15%) and OPEX (-50%) - Simpler (and then faster) system roll out - Lower energy consumption: up to 71% of traditional RAN Consumption. Sharing the processing and thus the power between different cell areas is a way to utilize the BS more effectively. - Easy extension of the network. The operator only needs to install new RRHs and connect them to the BBU pool to expand the network coverage or split the cell to improve capacity. If the network load grows, the operator only needs to upgrade the BBU pool s HW to accommodate the increased processing capacity. page 15
3- Advantages and benefits of Cloud RAN The advantages of Cloud RAN (China Mobile slides) are (2/2): (B. Haberland Slides, Base Stations in the Cloud, Sept 2012) - Operator Sharing: BBUs of different operators can share the RRH configuration, - More efficient implementation of M2M Communication in the wireless network, (China Mobile Res. Inst.) - CRAN is based on open platform, can support multiple standards, and smooth evolution. (Many other sources) - Allows macro cells to be easily complemented with a layer of small cells bringing added capacity to end users. - Great flexibility to allocate the processing resources according to the traffic evolution page 16 3.1- Associate with software-defined networks (SDN) Software-defined networks (SDN) Common themes : Source: Keynote by A. Goldsmith, VTC Spring, Dresde, June 2013 1- Separate control and data plane (Virtual Cell Concept, see [Salz12] / **add and GreenTouch Project) 2- Open and programmable 3- Vendor-agnostic (interoperable) 4- Network abstraction offered to applications on top of the network = ability to tailor network performance to applications. 5- Self-Organizing Networks (SON) Future wireless networks will divorce HW and SW, with commodotized HW and cloud SW to manage it (see architecture) page 17
3.1- Associate with software-defined networks (SDN) Proposed software-defined wireless networks (SDWN) Architecture Source Fig: Keynote by A. Goldsmith, VTC Spring, Dresde, June 2013 page 18 3.2- Associate with Active Antennas and beamforming Replacing previously-used antennas, the ultimate solution integrates RF components directly into the antenna. This is not just an RF head in the same housing as the passive antenna. An active antenna comprises many RF elements distributed inside an antenna array together with a common control unit that steers and shapes the beam. page 19
4- Normalization and commercialization Many C-RAN Trials in Commercial Networks in China for dense urban city, business street and campus see [1] China Mobile signed C-RAN MoU with Orange page 20 4- Normalization and commercialization NSN Liquid radio platform ([6]): Nokia Siemens Networks Liquid Radio, is a fluid network architecture and platform paving the way for software- defined networks. These platforms are linked by high-capacity, low-latency transport networks in a non-hierarchical, meshed architecture. They self-configure to use the available hardware, interconnects and air interfaces in the best way. The Intelligent Self- Organizing Networks (ison) suite adds the ultimate liquidity to multitechnology networks. page 21
4- Identified research issues (not exhaustive) 1- Reliability and resiliency: a cloud network may have more sources of failure than a physical counterpart. 2- Interoperability: standards, handover decision criterions and process between different RATs (Radio Access Technologies). One possible scheme is to use Software-defined Radio (SDR) to process any protocol ([4], NGMN). page 22 4- Identified research issues (not exhaustive) 3- RRM Issues and QoS Support : Cooperative Radio Resource Management for multi-cells. The multi-cell RRM problem has already been addressed in various academic studies (see [1]), using various optimization techniques in trying to determine the best RRM solutions (resource scheduling, power control, admission control) with regard to the throughput and with some specific constraints. Cooperative or multi-cell RRM in C-RAN has some specific considerations, mainly large numbers and complexity. [1] proposes to reduce the joint processing/scheduling to a number of cells in order to reduce the complexity incurred in the C-RAN network architecture. page 23
5- Identified research issues (not exhaustive) Radio resource sharing for hybrid cellular networks: design of cooperative radio algorithms that reduces interference between cells and improve overall indoor capacity. Source [4] NGMN, Jan 13 page 24 5- Identified research issues (not exhaustive) 4- Cooperative Transmission and Receiving: (from [4], NGMN) Joint processing and coordination techniques such as coordinated multipoint (CoMP) transmission and reception can be even more efficient with the centralized processing of C-RAN. Joint processing includes not only CoMP defined in 3GPP, but also many other technologies such as joint scheduling, joint interference alignment/ cancellation and other. page 25
5- Identified research issues (not exhaustive) (from [1]) 5- With virtual BBS or BBS pool: The RRHs signal can be routed to any one of the BBUs in the pool. Advantages: The dynamic carrier scheduling of resources within baseband pools enhances redundancy and then overall operational reliability of the baseband pool: when a baseband card or a carrier processing unit fails, the work load can be promptly redistributed to other available resources within the pool, and restore the normal operation. Load balance among BBUs according to dynamic network load Reduction of system power consumption. According to [1], a large scale baseband inter-connect solution should be able to support 10-1000 macro BS page 26 5- Identified research issues (not exhaustive) 6- Energy Efficient/Green Infrastructure issues in C-RAN ([1]): Firstly, with centralized processing of the C-RAN architecture, the number of BS sites can be reduced several folds. Thus the air conditioning and other site support equipment s power consumption can be largely reduced. In low-traffic periods, many virtual BSs (RRHs) may be served by one BBU. Then, many other can be shutdown, leading to a reduced BBU pool energy consumption (less cooling, cost, ). Yet, this creates a new topic to address: BBU migration for real-time services. When moving from one BBU to another a vbs (RRH) is silent and this represents an interruption of several seconds* see C. Wang, et al, A Study on Virtual BS Live Migration A Seamless and Lossless Mechanism for Virtual BS Migration, in PIMRC 2013 * According to E. Dahlman, 3G Evolution, 2008, The maximum allowed interruption time for LTE handover is 300 ms page 27
5- Identified research issues (not exhaustive) Energy Efficient/Green Infrastructure issues in C-RAN ([1]): In addition, because the BBU pool is a shared resource among a large number of virtual BS, it means a much higher utilization rate of processing resources and lower power consumption can be achieved. /** Ongoing: model to estimate the economy of the BSs page 28 5- Identified research issues (not exhaustive) 7- Adaptability to Non-uniform Traffic [1] C-RAN is very suitable for non-uniformly distributed traffic due to the loadbalancing capability in the distributed BBU pool. Though the serving RRH changes dynamically according to the movement of UEs, the serving BBU is still in the same BBU pool. As the coverage of a BBU pool is larger than the traditional BS, non-uniformly distributed traffic generated from UEs can be distributed in a virtual BS which sits in the same BBU pool. page 29
6- State of the art of research works 6.1- CMG Scheduling Wang, et al, Wireless Network Aware Cloud Scheduler for Scalable Cloud Mobile Gaming in ICC 2012 propose a Wireless Cloud Scheduler (WCS) in the context of several geographical regions with different coverage for either WiFi network, or a Cellular network, or both, with different bandwidth and delay characteristics, and a set of CMG servers each with a certain Cost CLOUD. Fig. Source: Wang, et al, ICC 2012 page 30 6- State of the art of research works 6.1- CMG Scheduling The WCS considers simultaneously : - the constraints of the wireless networks that may be available to each CMG user, - the cost of available cloud resources, while scheduling : - the most optimal wireless link, - the cloud server, for each CMG session. Innovation : adding the problems of cloud computation scheduling to the largelystudied wireless network access scheduling. page 31
6- State of the art of research works 6.2- Wireless Transmission on Optical Network Wireless Signal Transmission on Optical Network challenges are summarized in [1], section 4.1: BBU-RRH Bandwidth Requirement Transportation Latency, Jitter and Measurement Requirements Data compression techniques on the LTE/LTE-A BBU-RRH, OBRI (Open BBU- RRH Interface) BBU-RRH Optical Fiber Networks: in order to avoid having every RRH fully occupy two optical fibers on a physically routed pair, the RRHs can be connected to each other in a cascaded manner according to the CPRI/Ir/OBRI interface specification. This permits two different routing trunk cables to form a ring and be connected to the same BBU, as shown in the Figure (from [1]) page 32 7- Our Work? Scheduler considering the energy cost? Radio resource allocation issues: We intend to work on Coordinated Radio resource allocation and power allocation schemes for multi-cells. Specific LTE-A environment page 33
References [1] C. Chen, C-RAN: the Road Towards Green Radio Access Network, China Mobile Research Institute, White paper (same author, same title), October 2011 Slides, August 2012 [2] J. Chen, et al, Open Wireless System Cloud: An Architecture for Future Wireless Communications System, Network and Communication Technologies; Vol. 1, No. 2; 2012 [3] A. Goldsmith, Enabling the Wireless Cloud Through Software-Defined Networking, Vehicular Technology Conference, VTC Spring, Dresde, June 2013 [4] NGMN, Suggestions on potential solutions to C-RAN BY NGMN Alliance, Jan 2013 [5] Manzalini, Clouds of Virtual Machines in Edge Networks,» IEEE Com. Mag., July 2013 [6] NSN White Paper, «Liquid_Radio Let traffic waves flow most efficiently,» June2013 page 34