I have blogged about Cloud RAN or C-RAN in the Metrocells blog here and am looking forward to more discussions on this topic in the SON conference later this year.
I came across this interesting presentation from Orange in the LTE World Summit this year where the authors have detailed the C-RAN architecture and also discussing the fronthaul challenges faced by C-RAN. The presentation is embedded as follows. Please feel free to add your comments with your opinions.
See my paper on "base station application optimizer" (DCNET, July, 2010) at
ReplyDeletehttp://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=5740723
A very good presentation on CRAN and COMP. Different deployment scenarios are discussed and covered very well. This whole thing is based on the assumption that max RRU distributed with AAA antenna at the max. I still feel that later on transporting CPRI traffic on GPON/ PON, here may be some changes to add PHY and MAC on the RRU and hence reduce the real time traffic. This needs to be backed up by the SoC vendors and carriers. I am also curious to know, what will be the RRU price for indoor hybrid DAS/ Small Cell implementation.
ReplyDeleteAntennas with built in RRHs are already a reality. AIR antenna from Ericsson is an example. C-RAN will become a reality soon with CPRI and compressed CPRI interfaces.
ReplyDeleteC-RAN is predicated on having a lot of fibre, so those operators that already have it, or are commanded to have it, will make use of it. The business case is unproven, especially as most data use takes place indoors where it may be harder to place RRUs and get the necessary 'fronthaul'
ReplyDeleteInteresting post indeed, have also a look on KT's C-RAN network evolution strategy (WARP+) begining with a centralized approach of BBU and then incorporate down the road complete virtualization per 1,000 cells group...increases overall spectrum efficiency but hey it requires a bunch of dark fibers ;)
ReplyDeleteCPRI front haul traffic don't have a lot of margin for compression and based on my understanding probably 2:1 compression may be possible in coming years. Then also for a 20 MHz BW and 8 RH, you need 6 Gbps instead of 12 Gbps. Even this much bandwidth will be too difficult to transport by the carriers who don't own GPON/ PON networks. So I still feel even in long term, the CRAN deployment will be transport media limited. Ultimately you have to compare if the 150 MBps DL+75Mbps traffic in DL through a 20 MHz SFN small cell (BBU+RRU combined) backhaul is costlier than CRAN architecture 10 Gbps. I still feel there is a long way to go for CRAN for a mass deployment compare to standard HeNB or enterprise ENB architecture. Yes flexibility wise there are lots of advantages to CRAN like 3D-MIMO, eICIC, BBU pooling, Neutral Host and rapid deployment feature.
ReplyDeleteC-RAN provides a "perfect" architecture for the advanced SON (such as AirHop's eSON) because it makes coordination and real-time optimization much easier by centralizing the network elements.
ReplyDeleteHi Yan,
ReplyDeleteYes I do agree with you, Most of the performance counters are measured on PHY and MAC and they are the part of BBU pooling , so for you it is a centralized processing. Need to understand more how the CPRI compression techniques evolve. Another are is how EMS handles the O&M of RRU based on muxed CPRI on PON or GPON.
Ravi
Here is comparison of Line rate, latency and service distance for fronthaul and back haul. Taken from Ceragon white paper.
ReplyDeletehttp://sdrv.ms/13XoCN5
Have a look in this paper about C-RAN based on Radio-over-Fiber Technologies:
ReplyDeletehttp://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6564455&tag=1
What are your opinion about this?
Zahid, It is wonderful presentation for fronthaul concept. Here is another solution for fronthaul based on active WDM. http://asia.lteconference.com/wp-content/uploads/1140120903-flexiHaul_solution_for_4G_and_Beyond%28Down%29.pdf flexiHaul has been deployed in SKT, the largest LTE operator in Korea. All the deployments are based on ring topology. Welcome you to contact me if you need more info.
ReplyDeleteActus is providing a compression algorithm to RAN vendors. It is based on FPGA on both BBU and RRH. The compression ratios are 2:1 and 3:1.
ReplyDeleteIf you need detailed information, please send me email.
choho@hfrnet.com
Interesting and timely discussion.
ReplyDeleteRavi Sinha makes good points about the use being likely among operators with access to existing fiber networks. I don't view the issue of compression vs. bandwidth as being fundamental in driving CRAN and COMP forward. Moreover, I view this as part of the larger evolution of the ICT/WAN and as still being early in building upon the latent technologies exposed in 3GPP r9 and beyond.
From the various posts with links to additional presentations and white papers, we can see both widespread interest, notable progress, and also continued rapid evolution of concepts that are coming into commercial stage for deployments. That brings up questions whether the efforts are becoming fractured and the pros and cons of the multiple approaches.
If we look broadly an ICT development we see much overlap in contested approaches despite the common network architecture framework including software defined radio, software defined network, and self organizing network that can be scaled to user deployable station/nodes.
What is likely to evolve will be tethered by the CRAN-COMP approach, with the degree of use varying by geography, operator, degree of existing fiber deployment, cost of new deployment, market demands, availability of capital, and other factors. The HetNet as commonly deployed is an early-stage development that while revolutionary in a sense, is yet a harbinger of a layered approach to network architecture.
The exploration of new layers or tiers of the overall network will likely involve both CRAN-COMP, and distributed eSON (evolved Self Organized Network) architectures as non-exclusive and often complimentary developments.