Network Clusters: Understanding and Optimizing Their Potential
Introduction: The Fundamental Shift from Sites to Clusters
The huge impact that mobile devices have had on humanity and the growing hunger for mobile network capacity have put a strain on mobile network operators. Operators must keep pace with the demands of the market or face the prospect of losing market share. The problem is that the costs to maintain and upgrade their networks are skyrocketing while the revenues on core services are declining. To make matters worse, each new network expansion introduces another generation of equipment and another layer of technology that will complicate interoperability and reduce efficiency.
From an engineering perspective, CTOs of mobile network operators are finding themselves in a vicious cycle that has them continuously deploying more sites only to create operational complexity and additional costs when it is time to maintain and upgrade them.
That can change.
The answer to this problem lies in rethinking our approach to network management. The current school of thought is to consider performance improvements based on what can and can’t be done at the cell site level. In order to make the network stronger, (i.e. adding more resources for increased capacity), operators know that they must invest at the site level. Shouldn’t the same imperative be assigned to making the network smarter (achieve higher efficiency and focus on the optimization of existing resources)?
This is where the conundrum begins. A smarter network requires that units such as BBU, DU and CU, the “brains” of the operation, establish a clearer, more comprehensive view of network traffic, usage spikes, “dark spots” and interference patterns. That is hard to achieve when placed at the cell site level. In dense areas, in particular, the cell coverage areas are so small there is no chance they can identify from which adjacent cells does the traffic for voice and data usually build up?
Instead of dozens of individual sites operating autonomously, why not cluster them so you can coordinate and control them all together?
How do Network Clusters help?
The fundamental change of seeing the network through clusters, as opposed to the traditional orientation around sites, is to gain better overall visibility into network capacity and demand in order to load balance, prioritize and even share resources of multiple cell sites. The result is a network that can quickly adapt itself to changes in demand by efficiently allocating cluster resources. We call this Network Agility.
This can help to address the variability of mobile communication traffic and the heterogeneous capabilities of different sites and different network technologies. Ultimately, the result is better network efficiency (optimized investments) and a higher capability to deal with local and temporary increases in demand (better performance).
In the broader context, optimizing with network clusters can improve day-to-day network performance.
- Power Efficiency
- Using an aggregated “view” over dozens of cells, sites can be selectively powered up or down based on actual demand. This reduces energy costs without sacrificing the ability to scale up when needed.
- Spectrum Efficiency
- A broader view of traffic patterns allows a deeper understanding of the ways cells interfere with each other in their specific service areas. This helps prevent conflicts between users on the same spectrum. The result is the ability to accommodate more users instead without having to invest in additional spectrum.
- Network Utilization
- Clustered resources create a network that is adaptable, enabling dynamic network changes by efficiently allocating resources to where they are needed from cells where they are not. This will free up existing network capacity instead of adding new capacity that will be inefficiently utilized.
- Schedule Efficiency
- Clusters create a much larger inventory of “slots” available to the scheduler compared to a single site. This makes finding the right resources for each user (time, phase, spectrum) much easier and faster.
- Reduced Interference
- By mapping traffic patterns and intelligently allocating frequencies, the interference between sites can be reduced, leading to better layering and a smoother transmission.
- Improved Customer Experience
- By controlling multiple sites, phones can be steered to less congested cells, allowing for higher throughput.
- Decrease Dropped Calls
- By having more awareness of adjacent cells, the options for call handoffs can be expanded to reduce the number of dropped calls.
Network clusters lead to clustered resources
Each network cluster operates a policy manager that marshals all the resources (spectrum, wattage and beam radius) of the array of sites placed under its control. The sites pass traffic information upstream where the cluster can piece together a map of the network traffic across a broader region. With this broader awareness the cluster is able to devise and communicate policies downstream for the efficient use of the resources of all its sites. This can bring about huge improvements in power management, load management and network recovery.
Use Case: Simpler, more efficient operation
From an operational standpoint, deploying and maintaining the intelligent components at a cluster level is vastly more efficient. Rather than apply upgrades to each and every cell site, a number that is constantly growing, engineering teams can focus on the clusters. This means that network operations, by installing those components only at the cluster level, can focus their efforts by minimizing the potential points of intervention by 95%-99%.
Use Case: More efficiency, flexibility, resiliency
Leveraging the scale that comes from aggregating dozens of cell coverage areas, their BBUs, clustered network resources can be centrally managed. First and foremost this creates built-in failover capabilities in the event of equipment or local network failures. In addition, these increased resources can be applied when and where they are needed, creating an inherently more flexible network that can dynamically scale according to shifting and variable demand.
Use Case: Better user experience
While mobile networks have been around for some time, there are vast differences between the generations of equipment that have been deployed. This can create issues for mobile devices traveling through an area, trying to maintain service while getting handed off between individual cells. Two users in the same vicinity may arbitrarily experience a different quality of service based on the chain of cells their devices have been connected to.
When the network can cluster its resources, the series of handoffs of a device tracing a path through such a heterogenous area can be optimized. For example, voice calls can be offloaded to adjacent cells that are not flooded with voice traffic, regular, asynchronous data can be routed through less congested cells and larger streams of data or data that is marked with a higher priority can be routed through cells with the highest available capacity.
Use Case: Higher tolerance to usage spikes
Planning a network is challenging even when usage stays within normal levels. Ensuring that the network is capable of dealing with sudden spikes in demand is a nightmare. For example, a common challenge for mobile networks is dealing with spikes in traffic as a result of spontaneous crowds. Typically this generates heavily concentrated increases in volume of calls as well as time-sensitive messaging and large data uploads (images and videos).
CTOs who don’t have the luxury of over-engineering the network and wasting budget on state-of-the-art equipment that must be always ready but seldom used, will have to watch as the network becomes congested and thousands of mobile customers who are all connected to a handful of cells suffer service degradation at once. With clustered resources, the traffic can be steered in advance to adjacent cells all around the crowds, avoiding the degradation by optimizing the network resources.
What changes when shifting to a cluster-based network topology?
Adopting a cluster-based approach requires applying a different perspective to your network planning. While the technology is available, some components of the ecosystem should be re-evaluated.
- Transmission
- The transmission between cells and the cluster probably needs to be upgraded, to support lower latency and more throughput.
- Redundancy
- To ensure smooth functioning of the entire cluster at all times, it is important to design for both operational and geographical redundancy.
- Cluster-size
- In order to make the most of the capabilities of a cluster and to allow it to function effectively, careful thought should go into mapping the dimensions of each cluster. In order to apply coherent policies across the cluster footprint, it makes sense to align the clusters based on clear separation (geographical, logical or functional).
- Cell components
- The makeup of equipment and capacity of a cell should be determined according to the aggregated needs of the cluster to which it is assigned, instead of simply cloning all cells to have the same capabilities.
- Unique topologies
- For specific use cases, there are “special cells”. For example, an internal shopping mall may have a dedicated cell with a DAS (Distributed Antenna System). The handoff between such a cell and the adjacent, external cells is often anything but smooth. Placing all these cells in a single cluster can avoid these end-user disruptions.
The transition from a cell-based to a cluster-based network topology is an opportunity to rethink the way we have approached these, and other questions, to make sure that our networks are optimized according to the current realities of mobile communication.
Conclusion: Efficiency before Capacity
Optimizing network efficiency is important for MNOs to address far before they consider additional investments in capacity. This is a technical and financial imperative. As revenues on mobile services become leaner, the only way to ensure that operators will be able to see an adequate return on their investment in additional capacity will be when that capacity is utilized cost-effectively. Shifting to a topology of network clusters will unlock that efficiency by maximizing performance, reducing costs, and enhancing customer satisfaction.
In a centralized world, where network clusters allocate the resources to where they are needed from where they are not, operators can increase capacity without adding new cells. This will lead to a cascading effect of lower CAPEX, leading to lower OPEX, leading to higher profit margins.