⚡ electrical engineering

A Fairness-Constrained Potential Game Mechanism for Decentralized Microgrid Energy Trading

This paper proposes a fairness-constrained potential game mechanism for decentralized microgrid energy trading that embeds capacity-weighted fairness regularization into prosumer utility functions to guarantee Nash equilibrium existence and convergence while effectively protecting disadvantaged participants through an adjustable efficiency-fairness tradeoff.

Original authors: Jiaming Zhang, Guohao Luo, Xiaoxiao Wang, Xin Liu

Published 2026-08-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jiaming Zhang, Guohao Luo, Xiaoxiao Wang, Xin Liu

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the modern electrical grid, a quiet revolution is taking place. For decades, electricity flowed in one direction: from massive power plants to passive homes that simply consumed what was delivered. Today, however, the landscape is shifting. Rooftop solar panels, home batteries, and flexible appliances have transformed many households into "prosumers"—entities that both produce and consume energy. These neighbors can now trade electricity directly with one another, bypassing the traditional central utility. This peer-to-peer trading promises a more resilient and efficient local grid, where excess solar power generated by one house can instantly power a neighbor's air conditioner. Yet, this freedom comes with a hidden risk. In a market driven purely by self-interest, those with the most powerful equipment or the best locations tend to win the biggest rewards, while smaller or less fortunate participants are left with meager returns. If the little players feel they are losing out, they will simply stop participating, causing the entire local market to collapse. The challenge for engineers is to design a system that encourages everyone to trade freely while ensuring that the weakest participants are not squeezed out.

Researchers at the Beijing University of Chemical Technology have proposed a new way to solve this dilemma, treating the energy market not just as a financial exchange, but as a structured game where fairness is built into the rules. Their work focuses on a specific type of mathematical framework known as a potential game. In this context, a potential game is a system where every individual's decision to improve their own situation automatically pushes the entire group toward a better, more stable state. It is a mechanism that guarantees that if everyone acts in their own best interest, the group will eventually settle into a balanced equilibrium where no one has an incentive to change their strategy. The researchers asked a critical question: can this elegant mathematical structure survive if they inject a rule designed to protect the disadvantaged?

To answer this, the team developed a mechanism that embeds a fairness constraint directly into the decision-making process of each prosumer. Instead of waiting until the end of a trading day to redistribute money or judge who got a fair deal, the system adjusts the value of every trade in real time. If a participant with limited capacity—perhaps a home with a small solar panel and a modest battery—finds their earnings falling below a certain expected level, the system applies a mathematical penalty to the trading strategies of those who would otherwise profit at their expense. This penalty acts as a gentle nudge, encouraging the more powerful traders to adjust their offers so that the smaller players can still make a meaningful profit. Crucially, the researchers proved that even with this fairness rule added, the system retains its mathematical stability. They demonstrated that the game still possesses a "potential" structure, meaning that the decentralized trading process will naturally converge to a stable point without needing a central boss to force an agreement.

The team tested their idea through rigorous computer simulations using both synthetic data and real-world energy consumption records from the Ausgrid network in Australia. They created scenarios where some prosumers had very little capacity to trade, effectively simulating the "disadvantaged" players. In a standard, unregulated market where fairness was ignored, these small players earned almost nothing, with their minimum benefit dropping to a negligible 0.0506. However, when the researchers introduced their fairness mechanism with a moderate setting, the minimum benefit for these small players jumped to 0.5075. This tenfold increase showed that the system successfully protected the vulnerable participants without destroying the market entirely. The overall efficiency of the grid remained high, proving that fairness and economic performance could coexist.

Beyond the numbers, the study also addressed the messy reality of how computers communicate. In a real-world neighborhood, data packets can be delayed, or a node might temporarily lose connection. The researchers designed a "checked" asynchronous version of their algorithm to handle these glitches. In this version, a prosumer might calculate a new trading strategy based on slightly outdated information from a neighbor. Before accepting this new strategy, the system performs a quick check against the current reality to ensure it still offers a genuine improvement. The simulations showed that even with significant delays and occasional dropouts, the system consistently found a stable solution. In every test case, the algorithm converged to a state where no one could improve their situation by changing their mind alone, confirming that the mechanism is robust enough for real-world deployment.

The findings suggest that a fair energy market does not require a central authority to dictate prices or redistribute wealth after the fact. Instead, by carefully designing the incentives that guide individual decisions, it is possible to create a self-correcting system where fairness emerges naturally from the trading process itself. The researchers found that by tuning a single parameter—the weight given to fairness—they could adjust the balance between total economic efficiency and the protection of the weak. This offers a practical path forward for community microgrids, ensuring that the transition to decentralized energy does not leave the smallest players behind. The work confirms that it is possible to build a local energy economy that is not only efficient and stable but also fundamentally just, allowing every participant, regardless of their capacity, to find a sustainable place in the new energy landscape.

Technical Summary: A Fairness-Constrained Potential Game Mechanism for Decentralized Microgrid Energy Trading

1. Problem Statement

As distributed renewable energy, user-side storage, and flexible loads transform conventional consumers into "prosumers," Peer-to-Peer (P2P) energy trading has emerged as a viable alternative to centralized dispatch. However, purely self-interested trading mechanisms often concentrate benefits on high-capacity nodes, compressing the marginal gains of small-capacity or disadvantaged prosumers. This inequity threatens long-term participation incentives and market sustainability.

Existing approaches, such as Nash bargaining or benefit redistribution, often treat fairness as an ex post evaluation metric or rely on centralized coordination. A critical theoretical gap remains: when fairness objectives are directly embedded into individual utility functions to influence real-time decision-making, does the resulting decentralized game structure preserve the existence of equilibrium and the convergence of distributed algorithms? Specifically, dynamic updates of fairness references can introduce cross-agent dependencies that break the exact potential structure required for guaranteed convergence.

2. Methodology

The paper proposes a Fairness-Constrained Potential Game Mechanism designed to balance economic efficiency with the protection of disadvantaged participants while maintaining provable convergence properties.

System Model:

Topology: A community microgrid with $n$ prosumers connected via an undirected graph $G=(N, E)$ , allowing bilateral P2P trading.
Strategies: Each prosumer $i$ selects a discrete net trading quantity $q_i$ from a finite set $S_i$ .
Base Utility: The baseline utility $U^0_i$ includes private benefits (based on local generation/load), neighborhood coordination costs (penalizing supply-demand mismatches), and a community imbalance penalty.

Fairness Regularization:
To protect disadvantaged prosumers, the authors introduce a capacity-weighted fairness regularization term $R_i$ into the utility function:
$U^F_i = U^0_i + \gamma R_i$

Benefit Proxy: $x_i(q_i)$ measures the improvement in private benefit relative to a no-trading state.
Reference Signal: $\theta_i$ is a fixed fairness reference signal (e.g., historical mean or capacity-normalized benchmark) assumed constant during a single trading-solution epoch.
Penalty Term: $R_i = -w_i [\max\{0, \theta_i - x_i(q_i)\}]^2$ . If a prosumer's benefit falls below $\theta_i$ , a quadratic penalty is applied.
Weighting: The weight $w_i = 1/(C_i + \epsilon_c)$ is inversely proportional to capacity $C_i$ , ensuring smaller prosumers receive higher protection weights.
Trade-off: The parameter $\gamma$ controls the balance between efficiency and fairness.

Game Theoretic Formulation:

Exact Potential Game: The authors prove that under fixed fairness reference signals and finite strategy sets, the fairness-regularized game is an exact potential game. The global potential function $\Phi^F$ is the sum of the base potential and the sum of fairness terms.
Separability: By fixing $\theta_i$ within an epoch, the fairness term remains separable with respect to unilateral deviations, preventing the cross-dependence that would otherwise destroy the exact potential property.

Algorithms:

Fair Best-Response Dynamics (Fair-BRD): A synchronous algorithm where prosumers sequentially update strategies to strictly improve their fairness-aware utility.
Checked Async-Fair-BRD: An asynchronous variant designed for practical communication delays and node dropouts. It generates candidate actions based on outdated states but rejects any update that does not yield a strict utility improvement under the current true state. This "check" mechanism ensures the potential function never decreases.

3. Key Contributions

Utility Modeling: Development of a capacity-weighted fairness-regularized utility model where benefit protection for disadvantaged prosumers is integrated directly into individual decision-making processes.
Theoretical Proof: Demonstration that embedding fairness constraints into P2P trading utilities preserves the exact potential structure, provided fairness references are fixed during the convergence epoch. This guarantees the existence of a pure-strategy Nash Equilibrium (PSNE).
Algorithm Design: Construction of synchronous (Fair-BRD) and asynchronous (Checked Async-Fair-BRD) dynamics that are proven to converge to a PSNE in a finite number of steps via finite improvement paths.

4. Experimental Results

Numerical experiments were conducted using synthetic scenarios, stress tests, and real-world data from the Ausgrid dataset.

Convergence: Both Fair-BRD and the asynchronous variant converged to a Nash gap of zero in all tested scenarios, confirming the theoretical finite-step convergence. The potential function increased monotonically with every accepted update.
Efficiency-Fairness Trade-off:
- In stress scenarios (low-capacity prosumers), increasing the fairness weight $\gamma$ significantly improved the minimum and average benefits of disadvantaged users. For instance, raising $\gamma$ from 0 to 2.5 increased the minimum benefit of low-capacity prosumers from 0.0506 to 0.5075.
- The mechanism offers an adjustable trade-off: moderate fairness weights improve equity with only a marginal reduction in total system welfare.
Robustness: The Checked Async-Fair-BRD achieved a convergence rate of 1.0 under various communication delays (up to 5 steps) and node dropout probabilities (up to 0.2), though iteration counts increased with delay.
Real-Data Performance: Using Ausgrid data, the proposed Fair-BRD achieved system welfare levels comparable to centralized welfare local search (difference < 0.02) while maintaining a zero Nash gap and protecting low-capacity users. It outperformed baselines like Nash bargaining and capacity-based fair pricing, which exhibited larger Nash gaps or lower welfare.

5. Significance and Claims

The paper claims to provide a provably convergent and adjustable mechanism for decentralized microgrid energy trading. Its primary significance lies in resolving the tension between fairness integration and game-theoretic convergence.

Theoretical Insight: The work demonstrates that fairness does not inherently destroy the mathematical structure required for decentralized convergence, provided the fairness reference is treated as a fixed signal within a trading epoch. This establishes a unified framework linking fairness regularization, potential functions, and best-response convergence.
Practical Utility: The mechanism protects disadvantaged participants without requiring a central coordinator to enforce redistribution, making it suitable for autonomous, distributed microgrid operations.
Limitations: The authors note that the theoretical guarantees rely on finite strategy sets and fixed reference signals within an epoch. Global convergence across multiple epochs with dynamically updated references requires further analysis. Additionally, the exact potential property ensures equilibrium existence but does not guarantee uniqueness.

In summary, the paper presents a mechanism that mathematically embeds fairness into the self-interested decision-making of prosumers, ensuring that the pursuit of fairness does not compromise the stability or convergence of the decentralized trading market.