Home > Communication Service Provider Solutions > Enabling Telecom Transformation > RAN Pooling - The case for RAN pooling with Cloud/Virtualized RAN > Efficient RAN resource use with C-RAN pooling
Cloudification of RAN enables pooling of RAN NFs and resources for multiple cell sites in centralized locations. In a 7.2 LLS deployment, traffic to the RAN (between the cell site and RAN) varies as per the user load, per cell site. The statistical independence of this variance among cell sites leads to SMG at the central pool, which leads to better resource utilization when resources are pooled.
For example, Figure 4 demonstrates the benefits of pooling and SMG based on real life traffic of 42 cell sites over a one-week period [1]. The illustrative data is for 4G cells, as similar data for 5G deployments is unavailable at time of publication. The 42 cell sites cover both office and residential areas, including shopping areas, busy roads, and so on. Data usage at these sites fluctuates with the time of the day and across days of the week.
The blue line shows the resources required for a traditional RAN deployment without pooling, and hence without SMG. The orange line shows the RAN resources required with SMG. The green line shows the actual resource usage.
The more pooled RAN NFs, the better the overall resource utilization. A fully Centralized RAN pool permits maximum RAN resource sharing among cell sites, which leads to maximum SMG. The gain can be defined as the ratio of resources needed when RUs at cell sites are connected to the same number of CU and DUs, to the ratio of a variable number of vCUs and vDUs that are in use only when needed, as expressed by the equation below:
Looking at the data volume across the 42 cell sites and the fluctuation of traffic over time, Figure 4 corroborates that Centralized RAN with a pooled BBU delivers more than double SMG. This represents a more than 50% savings in resources with pooled Centralized RAN.
With a traditional D-RAN deployment, RAN resource planning needs to handle peak capacity of 6 Gbps per cell site. With a pooled centralized C-RAN deployment, fewer RAN resources are needed to handle peaks of approximately 2.5 Gbps – engineering and allocating RAN resources for the average needed capacity in the system, rather than for peak capacity at each cell site.
Figure 5 looks at data volumes at sector A, B, and C of each 3-sector cell site [1]. For Sector A (blue line), traffic tends to peak in the middle of the day. For Sector B (green line), traffic peaks in the morning and evening. Sector C (orange line) has traffic during the whole day and evening. In general, all sectors have very low traffic during the night, with usage peaks at different times of the day. In this scenario, resource pooling at different aggregation points, as seen in the Dual Split RAN deployment, would offer better resource utilization compared to traditional per cell site D-RAN, without sacrificing performance, capacity, or coverage.
Pooling RAN resources allows RAN NF workloads to be moved around or allocated to other sectors or cell sites, depending on the required capacity. At times of low traffic, the servers hosting the NFs can be freed from RAN NF workloads. The freed capacity (compute, memory, and other resources) can be used to run other non-RAN NF workloads like AI/ML models, data harvesting, and so on. C-RAN pooling enables dynamic scaling (up, down, reallocation) of server resources to meet the requirements of the overall RAN system or deployment, which traditional D-RAN systems cannot.