With the advent of 5G and the Internet of Things (IoT), the pressure on wireless networks is growing exponentially. But expansion is expensive and many operators are looking at options for open and virtualized RAN (ORAN, vRAN). At cell sites and data centers around the world, the RAN is being disaggregated into virtualized, cloud-optimized units. Rakuten has successfully launched ORAN while Dish has also committed to deploying ORAN. Telefonica Deutschland Deutschland has launched an ORAN pilot project and initial estimates are that the ability to select from a larger range of providers rather than dependency on traditional RAN vendors could cut radio network costs by 30 percent going forward.
As operators consider cloud-optimized vRAN and ORAN options, several obstacles emerge. Initial deployments are happening at cell sites and data centers and while compliant hardware solutions have evolved to support virtual and cloud deployments, existing offerings are not entirely open or optimized. While progress has been made, the journey is not complete.
Continuing the Open RAN Journey
Operators initially evolved legacy RAN solutions to become ORAN-compliant. However, the solutions remained proprietary and specific to each vendor. There were minimal accommodations to virtualized deployments and minimal interoperability with other solutions.
Later virtual RAN offerings incorporated X86 architecture and Intel’s FlexRAN™ framework yet continue to rely on specific RAN hardware. Instruction Set Architectures (ISA) and software development tools are open and available, however the framework that integrates the software with RAN hardware remains vendor specific.
The result is that neither approach is truly open, and the advantages of disaggregation and open source are lost.
Flexible RAN Functionality
In an ORAN, the Distributed Unit (DU) is responsible for real-time functions like scheduling and the Central Unit (CU) is responsible for non-real time functions and can control multiple DUs. The units are software-based and hosted in an Edge cloud data center or central office. Ideally, an ORAN runs CUs and DUs on off-the-shelf servers and is flexible enough to work with radio equipment from any vendor, but that isn’t happening yet.
A solution that takes full advantage of powerful RAN hardware combined with a flexible software framework that enables hardware portability without affecting latency or performance is the next step on the journey to an open and portable ORAN. Whereas open implies that ISAs and tools are widely available as open source, portability requires a framework that enables rapid integration with any variety of physical and virtual network components from numerous vendors. Abstracting individual software and hardware layers is not enough to ensure this flexibility.
Flexible RAN functionality requires a uniform model for underlying RAN hardware and sets the stage for portable RAN software. The two key elements of this model are the RAN abstraction layer and RAN Hypervisor. The RAN abstraction layer is a repository of high-performance RAN functional components. The RAN Hypervisor captures the compute details of the hardware platform including CPU, GPU, digital signal processor, accelerator and the execution cost function associated with each compute element.
Building a Flexible ORAN
A flexible ORAN architecture must be optimized for RAN workload and include a heterogeneous multi-core digital signal processor to enable maximum functionality in software. Hardware is needed that can support Hypervisor-driven virtualization at scale. Scalability is critical in any network, and an ORAN solution must easily scale up or down depending on demand. As processing demands increase, power consumption becomes a critical factor and the ability to fully utilize the underlying silicon is key.
A disruptive solution from Saankyha Labs using its elastic-RAN (e-RAN) processor establishes a new hardware platform and software framework that is truly open and portable. Based on the underlying hardware architecture and applying scheduling policies and constraints; one or more RAN functions can be executed on a variety of target computing hardware. Unlike traditional Hypervisors, the e-RAN Hypervisor is not tightly coupled to the underlying hardware platform. Since it only requires a description of the target computing hardware, it can create a schedule for a different platform.
The Saankya Labs e-RAN processor is a heterogeneous multi-core chip that provides the foundation for building a truly open RAN. The chip can implement all the ORAN splits and be deployed for data centers or cell sites.
Benefitting from a Truly Open RAN
By deploying a purpose-built virtualized ORAN solution, operators derive all the benefits of powerful RAN-specific hardware combined with a flexible software framework that ensures the hardware is portable. While the hardware and software elements of the Saankhya Labs solution can be deployed independently, there are significant benefits to a combined solution.
- Total Cost of Ownership – Lower power consumption and size reduce overall costs. The flexibility to match any hardware architecture with any hardware and software provider eliminates vendor lock-in, changing the equipment cost dynamics. The DU can be collocated with the RU or pooled at a remote data center depending on need in a specific market. Dynamic compute provisioning optimizes utilization of silicon and prevents over-provisioning of resources.
- Elasticity - Flexible configuration and the ability to scale up or down depending on site size means that operators only deploy what they need and can quickly scale up when required. Capacity is created by adding additional chips or additional cards making it ideal for cell sites or data centers of any size.
- Portability – The chipset is not tied to specific hardware. The RAN abstraction layer and the RAN Hypervisor provide the framework for portable RAN software development. Using this framework in a telco cloud, operators are not tied to a single hardware vendor and can move between vendors as needed.
“DISH is actively building the first cloud-native, ORAN-based 5G network in the U.S. As we challenge the status quo with this next-generation network, we believe openness is fundamental,” said Sidd Chenumolu, DISH VP of Technology Development. “This approach enables us to use multiple vendors across software, hardware and the underlying architecture, enriching our network with best-in-class services. Saankhya Labs shares our commitment to openness, particularly in their DU architecture, which will help facilitate a modern, truly open RAN.”
Current ORAN solutions aren’t nearly open enough, negating advantages that should be derived from off-the-shelf hardware and open solutions. The Saankhya Labs Open DU solution for ORAN is a flexible approach to virtualizing the RAN. By enabling multiple vendor solutions to co-exist, operators have the flexibility to implement a truly open RAN. The solution is optimal for both cell site and data center deployment and can be readily adapted to the dynamic nature of the network. As a result, operators can mix and match hardware and software vendors and are no longer tied to a single framework, architecture, provider or solution.