Cloud RAN Resource Reconfiguration Controller for Hybrid GPP Accelerator Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cloud Radio Access Network (RAN) architectures with General Purpose Processing (GPP) and hardware accelerators face challenges in resource reconfiguration, particularly in meeting real-time requirements and supporting live migration of virtual machines and multiple radio access technologies.
Innovation Solution
A method for resource reconfiguration in cloud RAN that involves a controller determining system status and resource usage conditions, identifying resources to be reconfigured, and sending reconfiguration indications to target devices, including virtual machines and processing modules, with feedback mechanisms to ensure successful reconfiguration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a pure general purpose processing (GPP) platform is used in cloud RAN, then device complexity is reduced, but real time performance requirements cannot be met
Solution Approach 1:
The patent combines GPP platform with hardware accelerator array to form a hybrid processing system. The GPP platform handles higher layer processing while the hardware accelerator array handles real-time physical layer processing, merging the advantages of both approaches to achieve both reduced complexity and real-time performance requirements.
Solution Approach 2:
The hardware accelerator array is designed with universal processing modules that can be dynamically configured to handle different radio access technologies and processing functions. This multi-functionality allows the same hardware infrastructure to support various services while maintaining real-time performance through dedicated hardware processing.
2Ease of manufacture
If existing resource reconfiguration solutions are used in cloud RAN with GPP and accelerator array, then implementation is simplified, but resource sharing and live migration capabilities are insufficient
Solution Approach 1:
The patent implements dynamic resource reconfiguration capabilities where the hardware accelerator array can be dynamically allocated, deallocated, and reconfigured based on real-time service requirements. Virtual machine live migration is enabled through dynamic resource allocation, allowing resources to be dynamically reassigned between different services and nodes without service interruption.
Solution Approach 2:
A resource management module acts as an intermediary between the GPP platform and the hardware accelerator array, coordinating resource allocation, migration, and reconfiguration operations. This intermediary layer simplifies the implementation by providing standardized interfaces while enabling sophisticated resource sharing and live migration capabilities.
Data Source
Figure 1~2
Figure 3~5
Figure 6~7
AI summary
The present disclosure provides a method for resource reconfiguration and the corresponding device. A method provided in a first aspect of the present disclosure comprises: at a controller for managing a baseband unit (BU) pool, determining system status and resource usage conditions of the BU pool; based on the system status and the resource usage conditions, identifying a resource that needs to be reconfigured, the resource being selected from a virtual machine running on a server in the BU pool or from a processing module in an accelerator in the BU pool; determining an address of a target device using the resource, wherein the target device is selected from servers or accelerators in the BU pool; and instructing the target device to reconfigure the resource. The solution of the present disclosure effects reconfiguration of all resources in cloud RAN.