Page 40 - SAMENA Trends - June-July 2025

REGIONAL & MEMBERS UPDATES SAMENA TRENDS

China Mobile and Huawei Win TM Forum's 2025

Excellence Award for Autonomous Networks

During DTW 2025, Huawei and China Mobile efficient operations & maintenance, high-value scenarios such as fault handling
won the Excellence Award for Autonomous which has effectively accelerated the and customer complaint resolution, the
Networks for the End-to-End Autonomous advancement of the telecom industry solution leverages telecom foundation
Network Operation Center (Dark NOC) towards level 4. To achieve the L4 industry model to build two main types of agents:
solution. This project focuses on high- goal of "end-to-end automation of NOC role-based Copilots and scenario-based
value Autonomous Networks scenarios, operations in high-value scenarios and Agents. This solution was first deployed
leveraging telecom foundation model self-service site operations", Huawei and in Guangdong and Zhejiang provinces,
and agents, and has achieved significant China Mobile have jointly created the End- achieving significant results including
achievements in end-to-end automation, to-End Autonomous Network Operation a 30% improvement in maintenance
quality & revenue enhancement, and Center (Dark NOC) solution. Focusing on efficiency and a 30% reduction in average
MTTR. Currently, the solution has been
commercially deployed across fault
management and complaint handling
scenarios in China Mobile Guangdong
and Zhejiang, covering mobile bearer,
wireless, core, and home broadband
networks. It is now being promoted to
other provincial subsidiaries, empowering
operators to serve tens of millions of
users. The successful implementation
of the End-to-End Autonomous Network
Operations Center (Dark NOC) Solution
provides a valuable practical reference for
global operators accelerating their journey
toward L4. In the future, Huawei and China
Mobile will continue to deepen innovation
and practical exploration in high-value
scenarios, injecting new impetus into the
automation and intelligent transformation
of the telecoms industry.

China Mobile Completes First Test of Remote Storage and Computing Via
HIC-OTN

To meet the high security requirements 240 km intelligent computing interconnec- in constructing large intelligent computing
for user data in foundation model training, tion network. This achievement marks a centers due to high costs and technical
China Mobile Research Institute recently significant milestone in the advancement requirements. Renting intelligent comput-
proposed a new remote storage and com- of intelligent computing center technolo- ing services can pose security risks when
puting architecture based on Hitless Intel- gies and service applications. The leapfrog transferring private data to external centers
ligent Computing-OTN (HIC-OTN). Working development of foundation model technol- for foundation model training. This creates
with China Mobile Hubei Branch and Hua- ogies is driving a surge in demand for intel- a significant gap between the pressing need
wei, CMRI has completed the industry's ligent transformation and upgrades across for AI capabilities and the scale of applica-
first technical test of remote storage and various industries. The computing power tion. China Mobile Research Institute in-
computing over 240 km using HIC-OTN in needed for training foundation models is troduced the cutting-edge HIC-OTN-based
China Mobile's intelligent computing cen- increasing; major tech companies, both in remote storage and computing technology
ter (Wuhan). This initiative establishes a and outside of China, are investing in clus- architecture. "Micro-computing power" is
new standard for secure foundation model ters with 10,000+ or even 100,000+ cards. utilized on the user side for data training,
training with localized user data. The foun- This involves high construction costs and with the training process segmented and
dation model, utilizing pipeline parallelism requires addressing technical challenges user data remaining stored locally. Only the
(PP) and with hundreds of billions of pa- to enhance the efficiency of large-scale intermediate values of model training are
rameters, achieved a training efficiency of computing power utilization. Small and sent through HIC-OTN to the service pro-
over 99% of that in a single cluster, on the medium-sized enterprises face challenges vider's intelligent computing center (pow-
40 JUNE-JULY 2025