Guangdong Mobile Embraces SPN Dynamic Energy Solution for Green Transport

Amid global warming, the Chinese government aims to achieve carbon peak by 2030 and carbon neutrality by 2060, as outlined in the 2021 State

Council Government Work Report. China Mobile has diligently implemented this directive by upgrading its “Green Action Plan” to the “C² Three Capabilities—Carbon Peak and Carbon Neutrality Action Plan” in 2021, setting clear energy conservation and carbon reduction goals. By the end of the 14th Five-Year Plan, comprehensive energy consumption and carbon emissions per unit of total telecom services are projected to decline by over 20% compared to the 13th Five-Year Plan.

The energy consumption proportion of China Mobile’s transport network ranges from 5% to 10%. There are obvious tidal effects and periodical traffic variations, but power consumption of network devices does not vary with service loads, leading to substantial wasted energy. To address these challenges, China Mobile Guangdong (Guangdong Mobile for short) and ZTE initiated AI energy-saving research and development for SPN in April 2021, successfully piloting the application in the existing Shaoguan network in May 2022.

SPN AI Dynamic Energy Saving Solution

The existing SPN experiences noticeable tidal effects and periodic traffic fluctuations over time. The innovative AI dynamic energy-saving solution collects real-time network data, analyzes and predicts changes in service load, and devises energy-saving schemes at chip, module, board, and network levels, ensuring intelligent power conservation while maintaining strict transmission performance.

Utilizing big data technologies and AI algorithms, the solution analyzes SPN traffic, predicts service trends, and implements precise energy-saving policies. It also incorporates various security protection mechanisms and energy-saving exit strategies according to network device and service operation statuses, thereby ensuring network and service security.

• Traffic Data Analysis: Determining Mainstream Scenarios

Based on half-year traffic data, scenarios are categorized into base station and private line traffic models by service type, high/low-traffic by traffic scale, regular/irregular-traffic areas by traffic features, and short/long-cycle scenarios by traffic changes.

• Traffic Trend Prediction and Energy Saving Policy Selection

Combining multi-model weights with the operational status of each period, long-term trend prediction adopts Prophet and LSTM algorithms to predict the next period’s traffic trend, generating periodic prediction baselines and determining maximum and minimum traffic prediction ranges.

The short-term early warning algorithm utilizes ARIMA and exponential regression algorithms to predict real-time traffic, assess trends, and make energy-saving decisions such as chip/module/board sleep based on real-time traffic trends. It evaluates the long-term trend model and dynamically adjusts it according to real-time traffic data.

• Energy Saving Safety Mechanism

In sandbox mode, the energy-saving algorithm simulates operation, evaluates its impact on devices and services, and calculates energy-saving effects.

• Multi-Layer Protection Mechanism

Given the fundamental role of the SPN in the communication network, ensuring network security is crucial. The energy-saving solution integrates a multi-layer security protection mechanism spanning from NE to network.

On the NE side, multiple regression algorithms are used for weighted analysis and prediction of network traffic to generate trend prediction models, build energy consumption prediction baselines, and calculate secure and reliable energy-saving spaces based on the traffic model. Moreover, the power consumption distribution inside the equipment is controlled according to its structural thermodynamic model. Combined with the design of ZTE’s in-house core chips, different components such as chips, fans, modules and boards within the equipment are adjusted to execute energy-saving policies in a secure and reliable environment.

On the network side, to deal with prevalent scenarios in transport networks like protection switching or traffic migration, the controller analyzes traffic relationships between NEs, calculates the protection path of each NE in real time, predicts potential traffic migration, and avoids burst step impacts that a single NE cannot predict. This approach establishes an energy-saving policy on the network layer, ensuring energy efficiency under stringent performance conditions.

• Energy Saving Exit Mechanism

To ensure uninterrupted service operation, the system incorporates three energy-saving exit mechanisms: when real-time traffic exceeds the preset threshold, a specific alarm is triggered, and the energy-saving mode is manually deactivated.

Verification in Guangdong Mobile’s Existing Network

The AI energy-saving solution was discussed multiple times by Guangdong Mobile and ZTE, underwent numerous lab verifications, and was deployed and tested extensively in China Mobile Shaoguan in May 2022.                                    

• Energy efficiency: If dynamic energy saving is disabled, the existing network of Guangdong Mobile with 955 sets of ZXCTN 6700 is estimated to consume 16.5177 million kWh. Enabling dynamic energy saving saves 15.28%, approximately 2.5239 million kWh annually, equivalent to 2.0191 million RMB in electricity costs.
• Security verification: The devices enter energy-saving mode to test four scenarios: burst traffic, board plugging/unplugging, tunnel switchover, and forced shutdown. No alarms or packet loss occur in the system, achieving the expected result. The system has operated securely and stably for over six months, enduring tests during major events such as Dragon Boat Festival, shopping festivals, college entrance examinations, National Day, and Spring Festival.
• Achievement promotion: Building on successful trials in Guangdong, China Mobile has actively promoted the SPN dynamic energy-saving solution to other provinces in the country. It’s now being applied in Heilongjiang, and will expand to Gansu, Guizhou, Zhejiang, Yunnan, and Fujian. Pilot projects and applications are expected to be completed across all provincial branches of China Mobile.