Cloud RAN – emergence in heterogeneity
While 3gpp standardisation efforts are devising ways to pack more and more bits into available spectrum within the capacity constraints governed by Shannon’s law, wireless radio network is on the move to create topologies that allow less and less number of users to share more and more of the available spectrum. There are two major macro trends that are splitting the network into completely opposite directions.
The first macro trend is pushing the network to deploy an underlay of 10’s of small cells per macro base station to add micro cells within a macro cell to initially improve coverage and then deliver capacity by serving fewer numbers of users. This trend simplifies radio access but creates complexity and scale challenges in the backhaul network.
The second macro trend is splitting the base station into a network. This trend simplifies backhaul but adds complexity to the radio access. Cloud RAN and high density base stations are other names of this trend. The move to distributed base stations in terms of remote radio heads has been happening long before the days of cloud.
Success of a distributed wireless network depends not only on a framework that allows adaptation and growth of distributed intelligence within the network via continuous self-learning but also on a solid foundation that allows coordination among the distributed intelligence to launch new services and profitable monetization of the network. Timing and synchronization is a key element of this foundation aside from self-healing, self-optimizing and software defined network functionality. Timing and synchronization dictates the network performance while software defined network eases effective network sharing and virtualization, service deployment and maintenance. Figure 1 shows a traditional base station architecture.
Figure 1: Traditional base station architecture. Click to enlarge.
Due to pros and cons of the above stated trends and ongoing deployments, there will be no clear winner resulting in coexistence and eventual coopetition of Cloud RAN technologies and small cells. However, it is author’s view that network evolution would favor the fundamental precept of successful network monetization that builds upon continuously growing intelligence within the network to launch new and better services. Given the proliferation of distributed base station and remote radio head technologies, it is likely that the wireless network evolution will tilt more towards cloud RAN in coming time (Figure 2). This path is not going to be easy. Operators and system vendors would need to work collectively in standardizing some critical elements of cloud RAN to bring about cohesive framework that eases adoption and guides a series of future innovations to this cause.
Figure 2: Distributed base station and remote radio heads. Click to enlarge.
In order to delve into standardization needs for timing and synchronization, it is important to understand the present state of the wireless infrastructure network. Base stations and associated access connectivity has been a closed network where both proprietary and pseudo-standards abound. However, there is a tremendous latent desire to leverage standards based connectivity and synchronization for performance optimization and economies of scale. The proliferation of Ethernet and slow but sure adoption of timing over packets is the right catalyst to fire this latent desire into an engine of innovations for the next generation cloud based network. As we all know that economically viable technology wins every day over better technology, care has to be given to create a framework that aptly serves the need for coexistence with the legacy, cost of deployment, and scalability concerns effectively. And this needs to be done without compromising delivering better and better network performance reliably and profitably.
Base stations synchronize with the core network by using combination of multiple timing and synchronization inputs. GPS and legacy TDM networks such as T1/E1 lines continue to be used along with new ways of synchronizing network nodes by packet timing protocol (PTP 1588v2) and synchronous Ethernet. PTP and synchronous Ethernet are gradually becoming mainstream ways in achieving synchronization within wireless infrastructure. The clock and control module (CCM) within base station leverages these different mechanisms to achieve frequency, phase, and time of day accuracy with the network. The base station clock control module in turn provides the synchronized clocking information to the radio elements. This is fairly straight forward in traditional centralized base station where radio elements reside in the same chassis. Successful distribution of synchronized clocks becomes challenging in distributed base stations where radio elements are located remotely.
In order to distribute timing and synchronization to remote radio heads, system vendors used proprietary protocols to begin with. Open base station architecture initiative (OBSAI) and common public radio interface (CPRI) standards were introduced to standardize connectivity between base station chassis and remote radio heads. CPRI and OBSAI protocols enabled transport of synchronization information along with essentially TDM based data plane transport. Stringent requirements of maintaining round trip deterministic latency to sub 16ns and timing alignment error for transmit diversity chains to within 65ns resulted in use of dedicated fiber links between base station chassis and remote radio heads.
In distributed base station network architecture, several remote radio heads are connected to the base station chassis using chain, tree, or star topologies. White CPRI and OBSAI standards support up to 40km long fibers, dedicated fiber for connectivity between each base station chassis and the radio heads is extremely limiting and expensive. For these reasons, majority of remote radio installations are constrained to distances of hundreds of meters from the base station chassis. A successful rollout of Cloud RAN based distributed base stations require fiber reach of up to 40km but more importantly over a shared network.
Critical first step to this lofty goal would be adoption of standard based connectivity and synchronization technologies in distributed base stations in order to set the stage for mainstream Cloud RAN deployments in 3-5 years’ time domain. Ethernet is going to be the obvious choice. The price points of 10G Ethernet ports are dropping quickly with ubiquitous deployments. In parallel, activities around standardizing timing and synchronization over Ethernet have picked up in standards bodies such as ITU and IEEE. Ethernet standards and rich ecosystem are ushering distributed base stations to transition to 10G Ethernet connectivity as an end to end protocol within distributed wireless network. However, a more focused effort within standards bodies is needed to address wireless application specific requirements for connecting base band chassis and remote radio heads or often termed as wireless infrastructure front haul. Support for mechanisms in standards to coexist with legacy pseudo-standard connectivity and synchronization schemes is critical to accelerate this transition.
Figure 3: Conceptual Cloud RAN network architecture sharing network using QoS/ traffic engineering. Click to enlarge.
Sharing of the network for front haul and widespread transition of distributed base station architecture to data center form factor leveraging virtualization is far out in time but underlying shifts are already beginning to happen in this direction. Challenges emerging from network latency, secure transport and stringent synchronization needs for coordinated multipoint transmission and reception will take some time to materialize to make shared network a reality. An engineered and traffic managed network with controlled access may be the first step to effectively address quality of service needs to connect bigger base band chassis to 10’s and even 100’s of remote radio heads. Sharing of this network among operators would be necessary to contain cost of deployments.
Coexistence of Cloud RAN and small cells in the race to ever growing capacity within networks is going to throw some scalability and interoperability challenges. Rolling in support for small cells in Cloud RAN architecture would be important to allow both the mega trends within wireless infrastructure to flourish in their domains. Cloud RAN initiatives are setup nicely to become a much needed network platform to abstract underlying heterogeneity enabling effective network monetization while easing network deployment and maintenance.
The author, Harpinder Matharu is the Senior Product Manager for the Comms Business Unit at Xilinx Inc.
www.xilinx.com
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
