By Joe Issac, Solutions Head, Network Engineering, Cyient
The concept of Open Radio Access Network (ORAN) has been under development for a considerable period. It was created with the goal of transforming the radio access network by introducing open interfaces and enhancing flexibility, thereby creating a vendor-neutral environment. Despite some successful greenfield trials, the full-scale commercial deployment of ORAN appears to be on hold. This delay is attributed to various factors, including the need for technology maturity, improved security measures, and enhanced energy efficiency.
Traditional Original Equipment Manufacturers (OEMs) in the Radio Access Network (RAN) market have traditionally restricted interoperability by tightly integrating their components, effectively leading to vendor lock-in for operators. This situation forces operators to rely exclusively on end-to-end equipment from a single vendor. Without open interoperability standards, entry barriers for new vendors into the RAN market remain high. In contrast, ORAN promises a more adaptable, cost-effective, and innovative approach to building and operating 5G networks.
Challenges in ORAN Hardware and Deployment
Within ORAN specifications, the gNodeB is divided into three key components: the Central Unit (CU), Distributed Unit (DU), and Radio Unit (RU), referred to as O-CU, O-DU, and O-RU. Further breaking down the CU into Control Plane (CP) and User Plane (UP) components allows for the deployment of diverse functionalities across different network locations and hardware platforms. However, managing the technical complexities and rapid technological advancements inherent in a multi-vendor component environment, while striving for tailored deployments through a concept known as “T-shirt sizing,” presents significant challenges in ORAN. T-shirt sizing involves custom designing ORAN software and network hardware infrastructure based on specific user or network requirements. Integrating these components effectively for pre-integrated deployments remains a challenging task for operators. Currently, most commercial ORAN deployments rely on hardware comprising selected OEM vendor products, often complemented with hardware accelerators to address performance limitations associated with x86 General Purpose Processors (GPP).
ORAN’s aim to enhance the Radio Frequency Physical (RF PHY) layer using a GPP hardware card faces challenges, as the existing GPP hardware falls short in addressing baseband processing performance issues. Unfortunately, plugging specialized AMD or ARM chipsets into GPP hardware for managing baseband data processing and performance isn’t a straightforward solution. The ORAN hardware architecture demands network and software separation to align with open architecture standards, but seamlessly transferring baseband processing between GPP, SPP hardware, and vice versa proves challenging. These limitations pose significant hurdles for operators looking to deploy ORAN while maintaining the separation of hardware and software.
On the other hand, the window of opportunity for 5G is slowly closing, and European telecommunications companies are increasingly considering the deployment of 5G Advanced or even 6G, given their challenging financial situations. Although organizations such as 3GPP, TIP, Open Air Interface, ORAN Alliance, and Open RAN Policy Coalition are actively addressing ORAN integration issues, progress has been slower than anticipated. Many telcos that have implemented ORAN are still tied to traditional OEM vendor solutions, remaining far from achieving the desired separation of hardware and software with extensive interoperability.
Streamlining Technology Processes in ORAN System Integration
ORAN specifications aim to simplify system integration by introducing plug-and-play-like deployment components. However, operators either need to integrate these components themselves or rely on skilled system integrators to customize and engineer ORAN solutions. The scarcity of skilled resources capable of designing and implementing ORAN solutions with interoperability in mind poses a significant challenge. An open RAN system revolves around a disaggregated, software-defined, and virtualized architecture using open interface specifications implemented in vendor-neutral GPP hardware. However, full-scale commercial deployment of 5G ORAN remains a distant goal. Adopting higher bandwidth and MIMO (Massive Input, Massive Output) technologies necessitates an efficient and enhanced Common Public Radio Interface (CPRI) interface with more functional splits to reduce overall front-haul (FH) bandwidth requirements.
One challenge with the evolved CPRI (eCPRI) specifications is that it lacks definition as an open and interoperable interface. Consequently, the Distributed Unit (DU) and Radio Unit (RU) must typically be sourced from the same supplier to ensure seamless integration. However, from an ORAN design perspective, an open front-haul interface enabling the ORAN Distributed Unit (o-DU) and ORAN Radio Unit (o-RU) should ideally be obtained from different vendors. Successful adoption and commercial deployment of open front-haul-based solutions hinge on bringing performance and capabilities on par with integrated solutions. Therefore, a disaggregated solution must support wide bandwidth, multi-band, and high output power requirements, low latency, low power consumption, and functional necessities such as network sharing and dynamic spectrum sharing.
Managing an Effective ORAN Hardware Rollout
Managing an effective rollout of ORAN hardware requires strict alignment among all RAN components to ensure seamless interoperability while adhering to standards and specifications. Even though ORAN specifications aim to simplify system integration, they ultimately drive toward achieving a plug-and-play deployment environment. Telecom operators must either undertake the integration process themselves or engage skilled system integrators to implement ORAN using a vendor-neutral hardware-based approach. The disaggregation of a hardware base station implies reaggregation with the necessary software, hardware, and separation of the plane ecosystem through virtualization, all while optimizing costs that telecom companies can afford. However, with full-scale commercial deployment of ORAN still a distant goal, telcos may find themselves relying on traditional vendor solutions or taking a cautious approach to ORAN deployment.
The O-RAN Alliance has made strides in developing interoperable interfaces between various parts of the network. However, these standards, while open, may require a combination of system engineering and integration expertise to design, develop, and integrate Open RAN subcomponents effectively. In the realm of Massive Input Massive Output (MIMO), where receiver (Rx) and transmitter (Tx) antennas are densely packed into the frame, performance improvements are observed when hardware and software are tightly coupled.
Technology Comparison
GPP vs. FPGA vs. ASIC Comparing different technologies, such as General Purpose Processors (GPPs), Field Programmable Gate Arrays (FPGAs), and Application-Specific Integrated Circuits (ASICs), reveals distinct advantages and trade-offs. GPPs like x86 processors are designed with high clock rates and excel at implementing a wide range of operations sequentially. They are particularly effective for solving general-purpose problems that match their instruction and data granularity. When dealing with problems that involve 32-bit data mapping well to the available instruction set and do not possess a significantly higher inherent parallelism than what the instruction set offers, a general-purpose CPU is an excellent choice.
In contrast, ASICs offer the advantage of precisely matching data types, instructions, and spatial parallelism to suit specific problems. While highly efficient for tailored applications, ASICs may be inefficient due to making worst-case assumptions, such as implementing the same number of processing elements regardless of the problem size. FPGAs inherit the advantages of ASICs but come at a cost factor, roughly 10 to 22 times higher. However, FPGAs can compensate for this cost differential by offering the ability to customize hardware to match the specific problem instance.
Toward a Hybrid ASIC-Powered Solution
The current use of cost-effective x86-based GPP hardware in open RAN deployments contributes to inefficiencies in open architecture RAN systems, resulting in performance-related issues. Given these challenges, we propose a hybrid hardware approach that combines ASICs and FPGAs as hardware accelerators to augment GPP hardware, particularly for enhancing baseband processing between RU and DU units. ASIC chipset hardware demonstrates superior performance in baseband processing and offers cost advantages compared to using GPP hardware or chip accelerators. ASICs can operate faster and more efficiently than GPP hardware or FPGAs, as they can be tailored precisely to baseband hardware processing tasks, eliminating resource wastage.
Radio Units (RUs) are typically implemented using FPGAs and ASIC boards, positioned in close proximity to RF antennas. Traditional FPGA accelerators are known for their power consumption, whereas an ASIC accelerator, built on modern technology platforms, consumes minimal power and may even operate solely on harvested energy for CU and DU white box hardware management. Leveraging intelligent hardware features can enhance radio performance, double network throughput, and reduce overall costs.