Page 365 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

assessment/optimization frameworks for the full lifecycle of network infrastructure, setting an
industry benchmark for sustainable intelligent infrastructure management for broader societal
benefits, including climate change mitigation.

Also, this project will extend its impact to empower green transformation across domestic and 4.3 - 5G
international industries, enriching AI + low-carbon applications in sectors ranging from smart
lighting and green buildings to campus commercial hubs, energy systems, and industrial
manufacturing.

3 Use Case Requirements

• REQ-01: Device Data Collection and Integration
It is expected that the infrastructure will support standardized data acquisition and
ingestion for devices within information infrastructure (such as data centers, base stations,
and equipment rooms), covering configuration, performance, energy consumption,
temperature [6], and service-level KPIs, both from the device side and the network
management (NMS) side [7]. The data collection and presentation follow industry
standards, ensuring high generalizability and interoperability.
• REQ-02: Device/Network Management Control Interface Specification
It is expected that energy-saving functionalities, instruction sets, parameters, and open
interfaces for various devices will be defined. Note: For example, base stations should
support symbol shutdown, channel shutdown, deep sleep, and extreme shutdown.
Cloud servers should support virtual machine migration, host power-off, and resource
downscaling. Air-cooled systems should support switching on and off, temperature
tuning, and adjustment of cooling capacity. Water-cooled systems should support startup
count control, frequency tuning, and temperature regulation. The platform must support
adaptation to standard and non-standard command protocols for heterogeneous devices
such as 4G/5G base stations, computing devices (CPU/GPU servers), and cooling systems
(air, water, or liquid cooling).
• REQ-03: Operational Requirements
It is expected that during energy-saving operations, the platform must monitor data
quality to ensure stable and accurate data streams. It should also continuously track the
stability of energy-saving device/environment KPIs, monitor alarm and energy-saving
metric changes, and provide mechanisms for manual intervention or emergency wake-up
in response to anomalies [5].

329