Computing platform architecture for secure and efficient deployment of cloud native communication network functionality

By using a combination of universal processors and dedicated accelerators in the telephone communication cloud infrastructure, the problem that the computing infrastructure cannot meet the high bandwidth and low latency requirements is solved, and high-performance, low-power network function deployment is achieved, suitable for 4G and 5G radio access network and cable/fiber broadband network.

CN118923101APending Publication Date: 2024-11-08GOOGLE LLC
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380029370.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-02-22
Filing Date
2023-02-22
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The prior art is difficult to meet high bandwidth, low latency, reliability and security requirements in integrated circuits of the same size, resulting in the imbalance of computing infrastructure cost-effectively and efficiently meeting business needs.

Method used

Provides a universal telephone communication cloud infrastructure, using a general-purpose processor based on streamlined or complex instruction sets and multiple dedicated accelerators, combined with integrated network interface cards, to build a modular computing platform architecture, supporting 4G and 5G radio access networks and virtualized deployment of cable/fiber broadband networks.

Benefits of technology

It realizes high-performance, low-power network function deployment, provides a flexible, secure and efficient platform that can meet artificial intelligence and security challenges, and improves the performance and efficiency of computing platforms.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118923101A_ABST
    Figure CN118923101A_ABST
Patent Text Reader

Abstract

The present disclosure provides a computing platform architecture for virtualization and cloud native network functionality. The architecture uses a general purpose processor based on a reduced instruction set computer and a plurality of dedicated accelerators and integrated network interface cards. Thus, the architecture can accommodate hundreds of gigabits of input / output.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of the filing date of U.S. Provisional Patent Application No. 63 / 312,662, filed on February 22, 2022, the disclosure of which is hereby incorporated herein by reference. Background Art

[0003] Communications service providers (CSPs) around the world are embracing disaggregation, cloud, automation and machine learning (ML) / artificial intelligence (AI) to become software-centric, agile and customer experience-centric. CSPs are virtualizing various network functions, leveraging cloud-native technologies in all areas of the end-to-end system architecture. The initial phase starts with the operations support systems and business support systems (OSS / BSS) that are typically deployed centrally in the CSP network, and in the later stages, virtualization extends to the core network at the regional data center or the CSP service edge.

[0004] Traffic on the Internet doubles almost every two years, and to maintain a proper balance between supply and demand, computing infrastructure also needs to double every two years. However, transistor density within the same size integrated circuit (IC), at the same power footprint and at the same cost is no longer doubling, which may create an imbalance where supply may not be able to meet traffic demand in a cost-effective and power-efficient manner. Summary of the invention

[0005] The present application relates to the deployment of virtualized / containerized network functions. One example relates to a virtualized distributed unit (vDU) of a 4G or 5G radio access network (RAN). The virtual distributed unit (vDU) network function of the 4G / 5G radio access network (RAN) involves the deployment of a data link layer including a physical layer, a scheduler, and control components of the data link. Considering the lower layer components involved in the protocol stack, the vDU places extremely stringent requirements on high-bandwidth computing: no packet loss, extremely low latency, predictability, reliability, and security. Some of these requirements require cloud infrastructure to deliver real-time performance. Wired access networks, such as cable modem termination systems (CMTS) in wired networks, may have similar system requirements. To handle such requirements in existing systems, vDUs are typically deployed on general-purpose processors (GPPs) together with backup acceleration building blocks to offload extremely high compute-intensive processing, such as forward error correction calculations. Incoming traffic in such an arrangement enters through a dedicated network interface controller (NIC), then a GPP-based central processing unit (CPU) that handles physical layer functions including fallback acceleration (Hi-PHY) to handle channel coding or forward error correction (FEC), and then again a GPP-based CPU that handles the scheduler and data link layer functions.

[0006] The present disclosure provides a universal and horizontal telephone communication (telco) cloud infrastructure that can lay the foundation for the virtualization of wireless networks such as 4G and 5G and other radio access networks (RAN) and wired access networks such as cable / fiber-based broadband networks. Such an infrastructure can be deployed in a highly distributed manner at hundreds of thousands of sites. Such an infrastructure can provide a flexible, secure and efficient platform to deploy all network and information technology (IT) functions in a seamless manner. Such an infrastructure can also provide higher performance and lower power consumption, while also bringing newer features to meet artificial intelligence and security challenges in the new world.

[0007] The computing platform architecture described herein provides secure and efficient deployment of CSP network functions, especially for access networks such as 4G and 5G RAN, cable and fiber broadband. The computing platform architecture can be modular, with a host computer as the main building block and an optional L1 processor as a PCIe device.

[0008] The present disclosure further provides a computing platform architecture for virtualization and cloud native network functions. The architecture uses a general-purpose processor based on a reduced or complex instruction set computer and multiple dedicated accelerators and an integrated network interface card. Therefore, the architecture can accommodate hundreds of gigabits of input / output. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 is a schematic diagram illustrating an example 5G deployment model.

[0010] Figure 2 is a block diagram illustrating an example framework that enables a cloud provider to service a 5G model in accordance with aspects of the present disclosure.

[0011] Figure 3 is a block diagram illustrating an example cloud platform architecture for cloud service provider network functions according to aspects of the present disclosure.

[0012] Figure 4A and Figure 4B Provided according to various aspects of the present disclosure Figure 3 Front and top views of an example of how a cloud platform is physically implemented in a server platform in a rack.

[0013] Figure 5 is a block diagram illustrating another example computing platform architecture for CSP network functionality in accordance with aspects of the present disclosure.

[0014] Figure 6 Shows how to combine Figure 5 An example of processing performed in architecture 500 is described. DETAILED DESCRIPTION

[0015] Figure 1 An example 5G deployment model is shown. The cloud platform 101 supports a hierarchy of sites, including central data centers 102, regional data centers 103, aggregation sites 104, pre-aggregation sites 105, cell sites 106, and in some cases enterprises 107. The number of central data centers 102 can be relatively small, while the number of regional data centers 103 is large. For example, the cloud platform 101 can support approximately 10 or fewer central data centers 102 and tens or dozens of regional data centers 103. Aggregations 104 can be in the order of hundreds, while pre-aggregations 105 can be in the order of thousands. Just by way of example, there may be a hundred or a few hundred aggregation sites 104, and a thousand or a few thousand pre-aggregation sites 105. Such a system service can have tens of thousands of cell sites 106.

[0016] In each of Models AD, automation, core, policy, and central services occur at the level of the cloud platform 101, central data center 102, and regional data center 103. In each of Models AC, the user planning function (UPF) and centralized unit (CU) are located at the aggregation 104. In Model A, the containerized distributed unit (DU) is located at the cell site 106. The cell site 106 includes a radio unit (RU) that can be used to establish a radio connection with a user device. In Model B, the DU is at the pre-aggregation 105 level. In each of Models A and B, the DU is a containerized or virtualized application, and the RU is a physical device. In Model C, both the RU and the DU are physical devices at the cell site 106 level.

[0017] In mode D, private 5G is provided to enterprise 107. Enterprise 107 can be, for example, a company or an organization. In this model, UPF, CU, DU are containerized or virtualized applications of the enterprise, and RU is the physical device of the enterprise.

[0018] Figure 2 is to show that cloud providers can serve 5G models with improved efficiency and security (such as combining Figure 12. Block diagram of an example framework of the discussed models A, B, and D). The framework includes a telecommunications analysis and assurance platform (TAAP) 210 in communication with a cloud edge platform 230. The cloud edge platform 230 may include a cloud management platform 231, a distributed cloud edge network engine 232, and a distributed cloud queue management engine 233. The edge platform 230 may further include a host operating system (OS) 234. An accelerator abstraction layer (AAL) 235 exists on top of the host OS 234. The AAL 235 may be controlled by the cloud platform or a third party. The edge platform 230 may further include a host CPU unit 236, which includes a packet processing accelerator and an ML accelerator. The host CPU 236 may execute an L1 physical (PHY) accelerator and PHY software 237. The PHY accelerator 237 may be controlled by a third party. The containerized DU application 220 may be controlled by a third party and communicatively coupled to the host CPU 236. As an example, the containerized DU application 220 may be a RAN of an independent software vendor (ISV).

[0019] Figure 3 Another example cloud platform architecture for cloud service provider network functions is shown. As shown in this example, a host computing unit 336 is coupled to an L1 accelerator 337 via a PCIe bus 380 .

[0020] The host computing unit 336 includes a host CPU 340 in communication with DRAM 352, storage 354, an edge tensor processing unit (TPU) 356 or other machine learning accelerator, processor or hardware unit, and a root of trust 358. The host CPU 340 is further in communication with a network I / O 362.

[0021] The host CPU 340 may be, for example, an application specific integrated circuit (ASIC) including multiple processing cores. For example, the host CPU 340 may include a NIC ASIC. The host CPU 340 may include any number of processing cores, such as 8, 16, 24, 32, 36, 48, 64, etc. According to other examples, the host CPU 340 may be any of a variety of other types of processing units, such as a graphics processing unit (GPU), a field programmable gate array (FPGA), a microprocessor, etc. The host CPU 340 may be implemented on a computing device, which itself may be part of a system of one or more devices. The host CPU 340 may include multiple processors that may operate in parallel.

[0022] DRAM 352 may be any type of dynamic random access memory, such as a DDR4 memory chip, etc. According to some examples, DRAM 352 may include multiple DRAM devices. Figure 3DRAM is shown in FIG, but other types of memory may be used in other examples. Such memory may store information accessible to the host CPU 340, including instructions executable by the host CPU 340, and data that the host CPU 340 may retrieve, manipulate, or store. Such memory may be a type of non-transitory computer-readable medium capable of storing information accessible by a processor, such as volatile memory and non-volatile memory.

[0023] Instructions may include one or more instructions that, when executed by a processor, cause one or more processors to perform actions defined by the instructions. Instructions may be stored in an object code format for direct processing by a processor, or in other formats, including interpretable scripts or a collection of independent source code modules that are interpreted or pre-compiled as needed.

[0024] Data can be retrieved, stored, or modified by a processor according to instructions. Data can be stored in a computer register, in a relational or non-relational database as a table with multiple different fields and records, or as a JSON, YAML, proto, or XML document. Data can also be formatted in a computer readable format, such as but not limited to binary values, ASCII, or Unicode. In addition, data can include information sufficient to identify relevant information, such as numbers, descriptive text, proprietary codes, pointers, references to data stored in other memories (including other network locations), or information used by functions to calculate related data.

[0025] The storage device 354 may include any type of non-transitory computer-readable medium capable of storing information, such as a hard drive, a solid-state drive, a tape drive, an optical storage device, a memory card, a ROM, a RAM, a DVD, a CD-ROM, a writable memory, and a read-only memory. For example, the storage device 354 may include a solid-state drive (SSD), a hard disk drive (HDD), a non-volatile memory express (NVMe), etc. According to some examples, the storage device 354 may include any combination of volatile and non-volatile memory.

[0026] The edge TPU 356 may be, for example, an ASIC designed to run AI at the edge of a cloud framework. According to other examples, the TPU 356 may be an FPGA, a general-purpose CPU, or other processing unit.

[0027] The root of trust 358 may be, for example, a hardware or software module that ensures that connected components can be trusted. For example, the root of trust 358 may be a security component that ensures that a device communicating with the host computer 336 has a valid certificate.

[0028] Network input / output (I / O) 362 may include any of a variety of I / O interfaces. For example, I / O 362 may include multiple different types of interfaces for communicating with different devices.

[0029] The host computing module 336 may operate in coordination with other components of the system, such as a voltage regulator 372, a cooling module 374, a power supply 376, a printed circuit board (PCB) 378, and the like.

[0030] The L1 accelerator 337 may also have an I / O interface. The L1 accelerator 337 may perform digital signal processing of a physical layer function of a network protocol stack. According to some examples, the accelerator 337 communicates with a global navigation satellite system (GNSS).

[0031] Figure 4A and Figure 4B A front view and a top view are provided, respectively, of a physical implementation of a server platform in a rack. As shown, Server 1 and Server 2 are powered by respective power supply units (PSUs) located near the respective servers in the rack. The rack also includes fans to provide cooling for Servers 1 and 2. Each of Servers 1 and Server 2 includes a respective PCIe accelerator. The PCIe accelerator may be, for example, Figure 3 L1 accelerator 337. Such an accelerator may be a third party component included in the server.

[0032] Figure 5 Another example computing platform architecture 500 for CSP network functions is shown. The architecture 500 provides secure and efficient deployment of CSP network functions, particularly for access networks such as 4G and 5G RAN, cable and fiber broadband. The computing platform architecture 500 can be modular, with a host computer 536 as the main building block and an optional L1 processor 537 as a PCIe device.

[0033] The PCIe L1 processor 537 may have an integrated network interface card (NIC) function 592 for integrated network input / output (IO); and a programmable, high-performance, and power-efficient Layer 1 packet and / or digital signal processor 594 that can handle all functions of the physical layer, allowing any GPP-based CPU to focus on the remaining tasks. Accessing network functions in the CSP network may have very strict latency and time-sensitive requirements. In order to provide high-precision timing synchronization, the PCIe L1 processor may also have a synchronization building block 596 on the module, which has related silicon structures such as a digital phase-locked loop (DPLL), a GNSS receiver, etc. The PCIe L1 processor 537 may, for example, be a software-based abstraction of the L1 processor. Such a software-based abstraction can make it easy for network function application developers to port from one hardware structure to another.

[0034] The host compute module 536 is the hub of the architecture, connecting itself to an optional PCIe L1 processor 537 via a plurality of PCIe lanes 580. Such PCIe lanes 580 may be Gen3, Gen4, Gen5, Gen6, and so on.

[0035] The host computing module 536 includes a processor 540. The processor 540 may be, for example, a next generation programmable and hybrid processor. By way of example only, the processor 540 may be a combination of a GPP CPU based on an energy-efficient 64-bit reduced instruction set computer (RISC) or complex instruction set computer (CISC) plus an integrated NIC with hundreds of gigabit network I / O plus multiple dedicated packet processors. The dedicated packet processors may enhance processing to achieve a good balance of flexibility, performance, power consumption, and cost.

[0036] An example of such a dedicated processor includes a bulk encryption accelerator 544, which uses IP security (IPSEC) to encrypt all services on all network I / Os in bulk at multiples of 100 gigabits. Another example of a dedicated processor includes a packet processing accelerator 542, which can perform packet processing and IP service forwarding at line rate. However, the architecture is not limited to these two examples and can have more functions in similar lines. The bulk encryption accelerator 544 can be used, for example, to encrypt / decrypt all network services from the system.

[0037] The processor 540 with integrated NIC, as well as specialized processors such as the packet processing accelerator 542 and bulk encryption accelerator 544, may be packaged in a system-on-chip (SoC) package for maximum performance and power efficiency.

[0038] The host computing module 536 also includes an onboard ML accelerator 556. The ML accelerator 556 can perform inference at the edge along with other functions such as DRAM 552, storage 554, and hardware root of trust 558 to enhance the trust / security of the disaggregated platform. The storage 554 can be onboard or reside on a separate physical device. The DRAM 552, storage 554, and root of trust 558 can be similar to the above combined Figure 3 DRAM 352, storage 354, and root of trust 358 are depicted. An optional GNSS receiver and time synchronization functionality 564 may also be present on the host computing module 536.

[0039] The architecture 500 may be implemented in any of a variety of forms of hardware. According to one example, the architecture 500 may be implemented as a system on a chip (SoC). According to other examples, the architecture 500 may be implemented in one or more servers in a rack. According to further examples, the architecture 500 may be implemented in any one or more computing devices.

[0040] Although multiple components of architecture 500 are shown, it should be understood that these are merely examples. Additional or fewer components may be included in other implementations, and components may be interchangeable with other types of components. Although some components are shown as being located within the same box, such components do not necessarily reside within the same physical enclosure.

[0041] Figure 6 Shows how to combine Figure 5 An example of processing performed in the depicted architecture 500. As shown, L1 processing may be performed at PCIe L1 processor 537. L2 and L3 processing may be performed at host compute module 536.

[0042] The computing platform architecture described above can be programmed for virtualization and cloud native network functions. In some examples, such functions can utilize components in the architecture, such as a 64-bit RISC or CISC-based GPP CPU and multiple dedicated accelerators and integrated NICs for multiple 100 Gigabit I / O. CSP network access features such as DU, CMTS, broadband network gateway (BNG), etc. can be densely deployed on the computing architecture. The architecture provides such deployment in a highly energy-efficient and high-performance manner.

[0043] Dedicated processors such as bulk crypto accelerators provide the ability to perform line-rate bulk cryptography on multiple 100 Gigabit IOs in the SoC package to protect all incoming and outgoing interfaces of CSP access network functions such as DU, CMTS, BNG, and core functions such as UPF or other security gateways. As an example, bulk cryptography can be performed using IPSEC. When deployed with a RISC-based GPP CPU, it can provide energy-efficient machine learning inference acceleration for CSP access network functions such as DU, CMTS, BNG.

[0044] The bulk crypto accelerator can be used to encrypt / decrypt all network traffic from the system. In some examples, the CNF software on the system can run with different L1 / L2 accelerators with minimal modification by using a hardware abstraction layer. In further examples, a cloud-based, intent-driven system securely and automatically manages the hardware and software on the compute modules.

[0045] The advantages of the above systems are that they improve performance and power efficiency and achieve efficient packaging of components. The architecture uses full inline acceleration, where the NIC is a bundled component of the processing complex. The system also has increased security. For example, security is improved by adding bulk inline encryption capabilities for all large incoming and outgoing traffic using IPSEC, adding fallback encryption for all control and management plane traffic using hardware accelerated SSL, and adding hardware root of trust to improve the integrity of the entire system (HW and SW).

[0046] Additionally, adopting machine learning at the edge enables the network to be self-driving, where ML inference is pervasive on the network edge. Hardware abstraction enables network function application code to be easily ported from one hardware implementation to another.

[0047] Unless otherwise stated, the foregoing alternative examples are not mutually exclusive, but may be implemented in various combinations to achieve unique advantages. Since these and other variations and combinations of the features discussed above can be utilized without departing from the subject matter defined by the claims, the description of the foregoing embodiments should be made in an illustrative manner rather than in a manner that limits the subject matter defined by the claims. In addition, the examples described herein and the provision of clauses expressed as "such as", "including", etc. should not be interpreted as limiting the subject matter of the claims to specific examples; rather, these examples are intended to illustrate only one of many possible embodiments. In addition, the same reference numerals in different figures may identify the same or similar elements.

Claims

1. A system, characterized in that: The system comprises: A host computing module, the host computing module comprising: one or more processors; one or more dedicated packet processors; and Integrated network input / output (I / O); and a peripheral interconnect processor, the peripheral interconnect processor being coupled to the host computing module via a peripheral interconnect bus; wherein the peripheral interconnect processor is programmable for L1 processing; and Wherein, the host computing module can be programmed for L2 and L3 processing.

2. The system according to claim 1, characterized in that The one or more specialized packet processors include a packet processing accelerator and a bulk encryption accelerator.

3. The system according to claim 1, characterized in that The one or more processors, one or more specialized packet processors, and integrated network I / O are implemented in a system on a chip (SoC).

4. The system according to claim 1, characterized in that The one or more processors include multiple Reduced Instruction Set Computer (RISC) or Complex Instruction Set Computer (CISC) processing cores.

5. The system according to claim 4, characterized in that The plurality of RISC or CISC processing cores include N×64-bit RISC or CISC processing cores.

6. The system according to claim 1, wherein: The integrated network I / O includes N×100G integrated network I / O.

7. The system according to claim 1, characterized in that The host computing module further includes a machine learning accelerator.

8. The system according to claim 1, wherein: The host computing module further includes a root of trust module.

9. The system of claim 1, wherein: The host computing module further includes a synchronization module.

10. The system according to claim 9, characterized in that The peripheral interconnect processor and the host computing module are coupled to a GPS receiver.

11. The system according to claim 1, characterized in that The peripheral interconnect processor includes an integrated network interface controller.

12. The system of claim 1, wherein: The peripheral interconnect processor includes a programmable system on chip (SoC).

13. A programmable computing platform architecture, characterized in that: The programmable computing platform architecture includes: General-purpose processor units based on reduced instruction set computers (RISC) or CISC; Multiple dedicated processors; and Integrated network interface card (NIC); Wherein, the RISC or CISC based general purpose processor unit, the plurality of special purpose processors and the integrated NIC are adapted to provide together at least 100 Gigabits of input / output.

14. The programmable computing platform architecture of claim 13, wherein: The architecture is provided in a system-on-chip (SoC) package.

15. The programmable computing platform architecture of claim 14, wherein: Further included is a communication service provider network function deployed on said RISC or CISC based general purpose processor.

16. The programmable computing platform architecture of claim 15, wherein: Further included is a software-based abstraction of the L1 processor.

17. The programmable computing platform architecture of claim 15, wherein: The architecture is suitable for protecting incoming and outgoing interfaces of a communication service provider access network function.

18. The programmable computing platform architecture of claim 17, wherein: Securing the incoming and outgoing interfaces includes performing bulk line-speed encryption.

19. The programmable computing platform architecture of claim 15, wherein: The architecture is adapted to protect the incoming and outgoing interfaces of the core functionality.

20. The programmable computing platform architecture of claim 15, wherein: The architecture is suitable for performing machine learning inference acceleration for communication service provider network functions.