TL;DR
This paper introduces a novel anti-jamming framework for the Internet of Battlefield Things (IoBT) by applying dynamic psychological game theory. By modeling the "intention to frustrate" between a soldier and an attacker, the authors demonstrate that accounting for human psychology can improve a soldier's mission success (material payoff) by up to 15.11% compared to traditional, purely rational game-theoretic models.
What Problem Does This Paper Address?
The Internet of Battlefield Things (IoBT) connects soldiers, sensors, and autonomous vehicles to provide situational awareness. However, these links are highly vulnerable to jamming attacks.
Traditional security models often rely on "perfectly rational" game theory (like the Nash Equilibrium), which assumes players only care about objective outcomes like data rates or power consumption. The authors argue this is insufficient for a battlefield. In reality, human combatants are driven by psychological factors—specifically, the desire to frustrate their opponent. A soldier isn't just trying to send data; they are trying to avoid the "anger" of a failed transmission they expected to succeed. Conversely, an attacker may prioritize jamming the "best" link specifically because it causes the most psychological damage to the soldier.
Methods
The authors formulate the interaction as a dynamic psychological game. The core innovations include:
- Psychological Utility Functions: Unlike standard games where utility is based only on actions, here utility includes a "frustration" term. Frustration is quantified as the positive gap between a player's expected payoff and their actual payoff.
- Second-Order Beliefs: To calculate frustration, players must form beliefs about their opponent's beliefs (e.g., "What does the attacker think I am expecting?").
- Bayesian Updating Algorithm: Because solving psychological equilibria analytically is computationally grueling, the authors propose a learning algorithm. Players use Bayesian updating to refine their first- and second-order beliefs over repeated interactions.
- $\epsilon$-like Psychological Self-Confirming Equilibrium (PSCE): This is the solution concept where players' beliefs eventually "converge" to match the actual strategies played in the game, even if they don't have a perfect global view of the opponent's mind.
Datasets
The paper does not use a real-world dataset. Instead, it employs a numerical simulation of a battlefield environment.
- Environment: A soldier moves from origin to destination along a path with $X$ (specifically 5 in the main simulations) uniformly deployed IoBT devices.
- Channel Model: Rayleigh fading with path loss (exponent $\lambda$).
- Parameters: Transmit powers ($P_S, P_A$) at 20 dBm, bandwidth at 20 MHz, and a tolerable delay threshold ($\Delta$) of 80 ms.
Key Insights and Ideas
- The "Frustration" Shift: In a standard Nash Equilibrium, an attacker might spread their jamming power to be efficient. In a Psychological Equilibrium (PE), the attacker becomes more prone to targeting the best available channel. Why? Because that is the link the soldier is most likely to rely on and expect to work; breaking it maximizes the soldier's frustration.
- Psychology as a Strategic Advantage: By anticipating the attacker's desire to frustrate them, the soldier can adopt a more robust "evasive" strategy. The paper shows that when the soldier correctly models the attacker's psychological intent, their objective material payoff (lower delay) actually improves.
- The Cost of Inaccuracy: The authors show that "learning to be rational" takes time. If a soldier acts on inaccurate psychological beliefs (e.g., after only 10 iterations of learning), they can suffer up to a 9.23% loss in payoff compared to the true equilibrium.
Key Results
- Performance Gain: Accounting for psychological considerations improves the soldier's material payoff by up to 15.11% over conventional dynamic games.
- Convergence: The proposed Bayesian learning algorithm successfully reaches an $\epsilon$-like PSCE, typically requiring about 210 iterations to converge.
- Sensitivity to Delay: As the soldier's tolerable transmission delay ($\Delta$) increases, the material payoff at the psychological equilibrium increases by up to 38.11%, as the soldier has more flexibility to evade the attacker's frustration-driven moves.
Limitations and Future Work
- Simplified Emotion: The model focuses exclusively on "frustration." Other battlefield psychological factors, such as fear, stress-induced cognitive bias, or overconfidence, are not yet incorporated.
- Static Jamming Power: The attacker's jamming power is assumed to be constant ($P_K$) when active. A more complex model could allow the attacker to vary power levels to further manipulate the soldier's expectations.
- Scalability: While the Bayesian approach is more practical than analytical solutions, the state space (histories) grows exponentially with the number of IoBT devices, which may require more advanced reinforcement learning techniques for very large networks.
Summary updated