Cloud RAN vs. Small Cells: Trading Processing for Transport Cost.
March 17, 2012
Perhaps you heard of Cloud-based RAN (C-RAN), the latest acronym in the mobile industry. If you are familiar with “base station hotel,” then you’re familiar with this mobile network architecture.
C-RAN takes the distributed base station architecture to its extreme: It pools the baseband processors in a central location and distributes remote radio headends which are connected through dark fiber to the baseband engines. The distance can be as long as 40 km.
The distributed base station architecture on which C-RAN is based gained traction through the second half of the 2000s as operators accepted placing radios outdoors. They traded the risk associated with placing active devices outdoors in exchange for considerable savings in energy and cost since placing the radio close to the antenna reduces transmit power requirements and consequently power consumption. The interface between the baseband and the remote radio has been standardized by a few industry consortia including CPRI (Ericsson) and OBSAI (Nokia).
But why stop there? Why keep the baseband at the base of the tower if it can be placed deep in the core? In areas where fiber is abundant this scenario would have benefits. Operators in Asia specifically are interested in C-RAN architecture because high densities of Asian cities along with high availability of fiber make the economics of C-RAN more palatable.
C-RAN is an architecture that directly competes with small cells which take the opposing approach. Small cells place the baseband together with the radio in a compact outdoor enclosure. In mobile communication, there’s no escape from distributing the radio modules. In this case, why not combine the baseband with the radio since the cost of deploying remote radio modules and compact base stations is all the same?
The difference comes down to the cost of transporting data between the core and the remote units, and it is very prominent. This is because the data between a compact base station and the core network has much lower bandwidth and is measured in the order of Mbps while the data between baseband and the remote radio is measured in the order of Gbps. In a compact base station, the backhauled data is actually user data plus some overhead for control and management functions of the base station. In comparison, the data between baseband and a remote radio is typically oversampled I/Q streams for different MIMO antennas. The higher the number of antennas in future base station, the higher the resulting data rate and bandwidth. The difference in transported data is huge and means C-RAN architecture is only feasible using fiber connectivity. Even though Ericsson recently announced CPRI over Microwave, wireless systems would not have sufficient capacity to handle some of the most advanced CPRI or OBSAI interfaces (OBSAI RP3.01 defines several rates including 3 and 6 Gbps).
C-RAN trades off the cost of processing against the cost of transport. I would venture to say that silicon (fundamental to the cost of processing) will always be cheaper than the cost of transport. Silicon can scale in ways that transport can never match. If fiber is the only way to transport small cells, then perhaps the economics would match. However, given the smaller bandwidth required to backhaul small cells, more options are available that can tilt the business case in favor of small cells. This is not to say that C-RAN has no applications. In fact, there are many possible use cases that make C-RAN attractive. I will address this in future posts.