With the rapid development of 5G technology, the importance of energy management for 5G base stations has become increasingly prominent. The high energy consumption of 5G base stations, driven by enhanced data transmission and coverage, poses significant challenges to operational efficiency and cost. This paper proposes a novel State of Charge (SoC) interval optimization method for 5G base stations, integrating demand response and battery energy storage system (BESS) coordination. The approach aims to optimize energy usage, reduce costs, and enhance grid interaction through a two-stage optimization framework. By considering communication load migration and renewable energy uncertainty, this method ensures reliable and economical operation of 5G networks.
The core of this research lies in developing a comprehensive model that accounts for the dynamic nature of 5G base stations. The model incorporates power consumption patterns, communication load transfer capabilities, and the degradation costs of energy storage batteries. The optimization framework is designed to handle uncertainties in renewable energy output and communication demand, making it robust for real-world applications. Through simulations, the proposed strategy demonstrates significant improvements in cost savings and energy efficiency compared to traditional methods.
In this paper, I will first outline the power consumption model of 5G base stations, detailing the components and their energy demands. Then, I will describe the communication load transfer model, which enables efficient resource allocation across different base stations. Next, the energy storage battery model is presented, focusing on SoC management and degradation costs. The two-stage optimization framework is explained, including the objective function and constraints. Finally, simulation results and analysis are provided to validate the effectiveness of the proposed approach.
Power Consumption Model of 5G Base Stations
5G base stations consist of two key systems: the power supply system and the communication system. The power supply system includes energy storage batteries, solar photovoltaic panels installed on rooftops, and grid connection points. The communication system comprises Active Antenna Units (AAUs), Baseband Units (BBUs), and network transmission facilities. The overall energy consumption of a 5G base station encompasses main station equipment, power supply systems, transmission devices, and monitoring and air conditioning systems.
The communication system’s energy consumption is primarily driven by AAUs and BBUs. Base stations can operate in active mode or sleep mode. In active mode, energy consumption is divided into dynamic and static components. Dynamic consumption is directly related to communication load, while static consumption includes base power. In sleep mode, energy consumption is minimized to standby levels. The power consumption model is expressed as follows:
$$P_{5G} = \begin{cases} P_{\text{Base}} + P_{\text{Active}}, & \text{active mode} \\ P_{\text{Sleep}}, & \text{sleep mode} \end{cases}$$
where \(P_{\text{Base}}\) represents the fixed power in active mode, including BBU, AAU, and other device power. \(P_{\text{Active}}\) denotes the incremental power related to the number of connected users, and \(P_{\text{Sleep}}\) is the standby power. Specifically:
$$P_{\text{Base}} = P_{\text{AAU}} + P_{\text{BBU}} + P_{\text{other}}$$
$$P_{\text{Active}} = \epsilon \cdot P_{\text{dy}}$$
Here, \(\epsilon\) is the energy efficiency coefficient, and \(P_{\text{dy}}\) is the additional transmission power due to user connections. This model allows for precise estimation of energy demands based on operational states.
Communication Load Transfer Model
Communication demands vary across different regions and times, such as residential, administrative, industrial, and commercial areas. To optimize energy usage, this paper proposes a communication load transfer strategy that adjusts connections between base stations and mobile users based on real-time load, photovoltaic output, and energy storage battery status. This facilitates efficient migration of information and data flows within the 5G network.
Define the set of 5G base stations as \(I = \{1, 2, \dots, i\}\), mobile users as \(J = \{1, 2, \dots, j\}\), and time periods as \(T = \{1, 2, \dots, t\}\). A binary variable \(c_{i,j,t}\) indicates the connection between base station \(i\) and user \(j\) at time \(t\), where 1 means connected and 0 otherwise. Constraints ensure each user is connected to only one base station per period:
$$\sum_{i \in S_{j,t}} c_{i,j,t} = 1$$
where \(S_{j,t}\) is the set of base stations accessible to user \(j\) at time \(t\). The transmission power for a connection is given by:
$$P_{\text{dy}, i,j,t} = (2^{\frac{B \cdot D_{j,t}}{10}} – 1) \cdot \frac{N_0}{10^{\frac{\alpha + \beta \cdot \lg(d_{i,j})}{10}}}$$
Here, \(N_0\) is noise power, \(B\) is channel bandwidth, \(D_{j,t}\) is user traffic demand, \(\alpha\) and \(\beta\) are channel fading coefficients, and \(d_{i,j}\) is the distance between base station \(i\) and user \(j\). The total dynamic power for base station \(i\) is:
$$P_{\text{dy}, i,t} = \sum_{j \in U_i} \alpha_{i,j,t} \cdot P_{\text{dy}, i,j,t}$$
where \(U_i\) is the set of users within the coverage of base station \(i\), and \(\alpha_{i,j,t}\) is a factor representing connection allocation. This model enables load balancing and energy savings through strategic user reassignment.
Energy Storage Battery Model
The energy storage battery system in 5G base stations is crucial for ensuring uninterrupted power supply and participating in grid demand response. Unlike conventional user-side storage, 5G base stations require reliable backup power to maintain communication services. With the adoption of lithium iron phosphate batteries, energy storage batteries can charge and discharge to interact with the grid, reducing energy costs while meeting demand. However, battery degradation costs must be considered, primarily influenced by State of Charge (SoC) and Depth of Discharge (DoD).
The operation of the energy storage battery is modeled as follows. The DoD is calculated as:
$$\text{DoD} = \frac{P_{i,t} \cdot \tau}{E_i^{\text{cap}}}$$
where \(P_{i,t}\) is the power output, \(\tau\) is the time interval, and \(E_i^{\text{cap}}\) is the rated capacity of the energy storage battery for base station \(i\). The cycle life \(L(\text{DoD})\) as a function of DoD is:
$$L(\text{DoD}) = A \cdot \text{DoD}^{-B} \cdot e^{-C \cdot \text{DoD}}, \quad A, B, C > 0$$
Here, \(A\), \(B\), and \(C\) are coefficients specific to the battery technology. The degradation cost per cycle is:
$$C_{\text{deg}, i,t} = \frac{C_i^B \cdot P_{i,t} \cdot \tau}{2 \cdot L(\text{DoD}) \cdot E_i^{\text{cap}} \cdot (\eta^{\text{ch}} + \eta^{\text{dis}}) \cdot \text{DoD}}$$
where \(C_i^B\) is the replacement cost of the energy storage battery, and \(\eta^{\text{ch}}\) and \(\eta^{\text{dis}}\) are charging and discharging efficiencies, respectively. To ensure reliability, minimum backup capacity is required based on regional grid reliability and communication load intensity:
$$E_{i,t}^{\text{re,min}} = \int_{t}^{t+T_i^{\text{re}}} P_{i,t} \, dt$$
where \(E_{i,t}^{\text{re,min}}\) is the minimum backup capacity, and \(T_i^{\text{re}}\) is the required backup time for base station \(i\). The battery operation constraints include:
$$0 \leq \beta_{i,t}^{\text{ch}} + \beta_{i,t}^{\text{dis}} \leq 1$$
$$P_{i,t}^B = \eta^{\text{ch}} \cdot P_{i,t}^{\text{ch}} – \frac{P_{i,t}^{\text{dis}}}{\eta^{\text{dis}}}$$
$$0 \leq P_{i,t}^{\text{ch}} \leq \beta_{i,t}^{\text{ch}} \cdot P_{i,\max}^{\text{ch}}$$
$$0 \leq P_{i,t}^{\text{dis}} \leq \beta_{i,t}^{\text{dis}} \cdot P_{i,\max}^{\text{dis}}$$
$$E_{i,t}^B = E_{i,t-1}^B + P_{i,t}^B \cdot \tau$$
$$\text{SoC}_{i,t}^{\text{low}} \leq \frac{E_{i,t}^B}{E_i^{\text{cap}}} \leq \text{SoC}_{i,t}^{\text{up}}$$
$$\text{SoC}_{\min} \leq \frac{E_{i,t}^B}{E_i^{\text{cap}}} \leq \text{SoC}_{\max}$$
$$E_{i,t}^B \geq E_{i,t}^{\text{re,min}}$$
$$E_{i,0}^B = E_{i,24}^B$$
$$b_{i,t} = \beta_{i,t}^{\text{ch}} + \beta_{i,t}^{\text{dis}} – \beta_{i,t-1}^{\text{ch}} – \beta_{i,t-1}^{\text{dis}}$$
These constraints manage charging and discharging states, power limits, SoC bounds, backup requirements, and cycle consistency. The cumulative degradation cost is:
$$C_{\text{deg}, i,t}^{\text{BESS}} = C_{\text{deg}} \left( \frac{E_{i,t}^{B,c}}{E_i^{\text{cap}}} \right) – C_{\text{deg}} \left( \frac{E_{i,t-1}^{B,c}}{E_i^{\text{cap}}} \right)$$
where \(E_{i,t}^{B,c}\) is the cumulative energy before state changes. This comprehensive model ensures efficient and cost-effective use of the energy storage battery.
Two-Stage Optimization Scheduling Model
The optimization aims to minimize the total operational cost of the 5G base station network, considering photovoltaic generation costs, energy storage battery degradation costs, and grid electricity purchase costs. The objective function is:
$$\min F_{\text{total}} = C_{\text{PV}} + C_{\text{BESS}} + C_{\text{grid}}$$
where:
$$C_{\text{PV}} = \sum_{t \in T} \sum_{i \in I} C_{\text{OM}, i}^{\text{PV}} \cdot P_{i,t}^{\text{PV}}$$
$$C_{\text{BESS}} = \sum_{t \in T} \sum_{i \in I} C_{\text{deg}, i,t}^{\text{BESS}}$$
$$C_{\text{grid}} = \sum_{t \in T} \sum_{i \in I} \omega_t \cdot P_{i,t}^{\text{grid}}$$
Here, \(C_{\text{OM}, i}^{\text{PV}}\) is the operation and maintenance cost for photovoltaic panels, \(P_{i,t}^{\text{PV}}\) is photovoltaic output, \(\omega_t\) is the time-of-use electricity price, and \(P_{i,t}^{\text{grid}}\) is the power exchanged with the grid. Power balance is ensured by:
$$P_{i,t}^{\text{grid}} = P_{i,t}^{5G} + P_{i,t}^B – P_{i,t}^{\text{PV}}$$
Additional constraints include data transmission limits:
$$G_{i,t} = \sum_{j \in J} \alpha_{i,j,t} \cdot G_j$$
$$G_{i,t} \leq G_i^{\max}$$
$$B_{i,t} = \sum_{j \in J} \alpha_{i,j,t} \cdot B_j$$
$$B_{i,t} \leq B^{\max}$$
$$P_{\text{dy}, i,t} \leq P_i^{\max}$$
To handle photovoltaic output uncertainty, interval prediction is used:
$$\underline{P}_{i,t}^{\text{PV}} \leq P_{i,t}^{\text{PV}} \leq \overline{P}_{i,t}^{\text{PV}}$$
where \(\underline{P}_{i,t}^{\text{PV}}\) and \(\overline{P}_{i,t}^{\text{PV}}\) are the lower and upper bounds of the forecasted photovoltaic output.
The two-stage optimization framework is designed to address uncertainties. In the first stage, the SoC status is optimized over the entire day, considering communication load migration and photovoltaic predictions. In the second stage, hourly BESS scheduling is performed within the optimized SoC limits to track real-time uncertainties. The problem is decomposed into a master problem and subproblems for efficient solving.
The master problem is formulated as:
$$\min_{x, y} F_{\text{pes}}^* + \min (x, y)$$
$$\text{s.t. } g(x, y, u_{\text{pes}}^*) = 0, \quad h(x, y, u_{\text{pes}}^*) \leq 0$$
where \(x\) represents first-stage connection decisions \(c_{i,j,t}\), \(y\) represents second-stage BESS scheduling variables, and \(u\) represents uncertain photovoltaic output \(P_{i,t}^{\text{PV}}\). The subproblems include optimistic and pessimistic scenarios:
$$\text{Optimistic: } \min_{u} F(x_M^*, y_M^*, u)$$
$$\text{Pessimistic: } \max_{u} F(x_M^*, y_M^*, u)$$
$$\text{s.t. } g(x_M^*, y_M^*, u) = 0, \quad h(x_M^*, y_M^*, u) \leq 0$$
These subproblems find the optimistic \(u_{\text{opt}}^*\) and pessimistic \(u_{\text{pes}}^*\) outcomes within the forecast intervals. The master and subproblems are solved iteratively until convergence. Then, the SoC interval optimization problem is solved:
$$\min_{y_1, y_2} \left[ F_P(x_{k+1}^*, y_1, u_{\text{opt}, k}^*) + F_P(x_{k+2}^*, y_2, u_{\text{pes}, k}^*) \right]$$
subject to constraints for both scenarios. Additional constraints ensure sufficient discharge capacity and SoC interval width:
$$(\text{SoC}_{i,t}^{\text{low}} – \text{SoC}_{i,t-1}^{\text{low}}) \cdot E_i^{\text{cap}} \geq P_{i,t}^{\text{dis}} \cdot \tau$$
$$\text{SoC}_{i,t}^{\text{up}} – \text{SoC}_{i,t}^{\text{low}} \geq \zeta_{i,t}$$
where \(\zeta_{i,t}\) is a design coefficient for the minimum reserved charging or discharging capacity. This framework enables adaptive and robust energy management.
Simulation Results and Analysis
To evaluate the proposed algorithm, simulations were conducted for four types of 5G base station regions: administrative, industrial, commercial, and residential. These regions differ in photovoltaic generation capacity and communication demands. The area considered is 2 km × 2 km, divided into four 1 km × 1 km subregions, with a base station density of 5 per km² and a coverage radius of 750 m. The day is divided into 24 time periods to simulate varying user traffic.
Key parameters for 5G base stations are summarized in the following table:
| Parameter | Symbol | Value |
|---|---|---|
| Energy Efficiency Coefficient | \(\epsilon\) | 2.857 |
| Max Transmission Power | \(P_i^{\max}\) | 800 W |
| Channel Fading Coefficient \(\alpha\) | \(\alpha\) | 37.6 |
| Channel Fading Coefficient \(\beta\) | \(\beta\) | 128.1 |
| Max Traffic Handling | \(G_i^{\max}\) | 410 Mbps |
| Max Bandwidth | \(B^{\max}\) | 100 MHz |
| Sleep Mode Power | \(P_{\text{Sleep}}\) | 500 W |
| Base Power | \(P_{\text{Base}}\) | 2300 W |
The energy storage battery backup behavior for each region is analyzed. Charging and discharging activities are sparse, with infrequent state changes. Before 16:00, base stations tend to charge the energy storage battery due to low electricity prices or high photovoltaic output, preparing for afternoon communication peaks. After 16:00, discharging occurs to reduce operational costs. Industrial and commercial areas, with higher photovoltaic output, can store excess energy, while administrative and residential areas rely more on grid power during low-price periods. Communication load migration shifts loads to areas with higher photovoltaic generation, improving overall energy efficiency. At night, base stations in administrative and commercial areas can enter sleep mode through load migration, further cutting costs.
To validate the SoC interval optimization method, four algorithms are compared:
- Algorithm 1: Baseline without communication load migration or SoC optimization, using hourly scheduling based on expected values.
- Algorithm 2: Includes communication load migration with dynamic energy storage backup.
- Algorithm 3: Adds SoC interval optimization with fixed interval width in a two-stage approach.
- Algorithm 4: The proposed method with dynamic SoC interval adjustment and real-time tracking compensation.
Using Monte Carlo simulations with 100 scenarios generated from normal distributions, the results are as follows:
| Algorithm | Average Cost | Cost Interval | Average Base Station Power | Average Grid Purchase | Average Grid Cost |
|---|---|---|---|---|---|
| Algorithm 1 | 921.3 | [882.8, 962.2] | 1620.4 W | 987.3 kWh | 809.6 |
| Algorithm 2 | 868.5 | [843.2, 891.3] | 1438.2 W | 925.2 kWh | 742.9 |
| Algorithm 3 | 782.4 | [751.5, 816.8] | 1313.6 W | 732.5 kWh | 572.8 |
| Algorithm 4 | 758.1 | [723.7, 791.9] | 1302.7 W | 693.6 kWh | 538.2 |
Algorithm 1, without communication load migration, requires more from the energy storage battery, reducing grid interaction and economic benefits. Algorithm 2, with load migration and base station sleep, lowers average power and cost. Algorithm 3 and Algorithm 4, with SoC optimization, show improved economics by adjusting output within SoC limits. Algorithm 4 outperforms Algorithm 3 due to dynamic interval adjustment, validating the proposed strategy’s effectiveness.
Conclusion
This paper presents an SoC interval optimization method for 5G base stations, integrating demand response and battery energy storage system coordination. By developing models for power consumption, communication load transfer, and energy storage battery degradation, and proposing a two-stage optimization framework, the approach effectively handles uncertainties in renewable energy and communication demand. Simulations demonstrate significant cost savings and enhanced energy efficiency compared to traditional methods. The dynamic adjustment of SoC intervals allows for flexible and economical operation, contributing to sustainable 5G network management. Future work could explore integration with wider smart grid applications and advanced battery technologies to further optimize performance.
The energy storage battery plays a pivotal role in this optimization, enabling reliable power backup and grid interaction. Through careful management of SoC and degradation costs, the proposed method ensures long-term viability and cost-effectiveness. As 5G networks expand, such energy management strategies will become increasingly important for reducing operational expenses and environmental impact. This research provides a foundation for future innovations in 5G base station energy systems, highlighting the critical importance of adaptive energy storage solutions.