Nash Equilibrium

The defining property of a Nash Equilibrium is strategic stability. Once the network of cognitive radios converges to this state, no individual secondary user has a rational incentive to deviate. If a radio unilaterally increases its transmit power to boost its own signal-to-noise ratio, the resulting co-channel interference it causes to others will degrade their performance, potentially triggering a retaliatory power race that ultimately harms the deviator's own objective function. The equilibrium holds precisely because any single change yields a negative net payoff.

Cognitive radios reach a Nash Equilibrium through iterative, distributed algorithms without a central controller. Common mechanisms include:

• Best Response Dynamics: Each radio myopically updates its strategy to maximize its utility given the current interference environment.
• Gradient Play: Radios adjust parameters incrementally in the direction of the steepest utility increase.
• Fictitious Play: Radios form beliefs about opponents' strategies based on historical frequency of actions. These algorithms are essential for Dynamic Spectrum Access (DSA) in infrastructure-less ad-hoc networks.

The existence and efficiency of a Nash Equilibrium are dictated by the utility function each radio seeks to maximize. A well-designed utility must balance conflicting objectives:

• Maximizing individual throughput or signal-to-interference-plus-noise ratio (SINR).
• Minimizing battery consumption via Transmit Power Control (TPC).
• Satisfying a minimum Quality of Service (QoS) constraint. A common formulation is a sigmoidal function where utility increases with SINR up to a saturation point, preventing radios from wasting power for marginal gains.

The Price of Anarchy (PoA) quantifies the efficiency loss when self-interested radios operate at a Nash Equilibrium compared to a globally optimal, centrally planned allocation. In spectrum sharing, a high PoA indicates that non-cooperative behavior leads to significant tragedy of the commons, where aggregate interference reduces total network throughput. Cognitive radio designers combat this by introducing pricing mechanisms or virtual currency into the utility function, effectively taxing radios for the interference they cause to align individual incentives with social welfare.

A Nash Equilibrium is not guaranteed to exist or be unique in every spectrum sharing game. Key mathematical conditions for a pure-strategy equilibrium include:

• Quasi-concavity of the utility function in the radio's own action.
• Compact and convex strategy sets (e.g., bounded transmit power ranges).
• The game must be a potential game or a supermodular game for standard learning algorithms to converge. If these conditions are violated, radios may oscillate indefinitely between strategies, a phenomenon known as a limit cycle, causing catastrophic network instability.

Beyond pure strategies, cognitive radios can employ randomized strategies:

• Mixed Strategy Nash Equilibrium: A radio probabilistically selects from multiple pure strategies (e.g., choosing a channel with a certain probability) to make its behavior unpredictable to competitors. This is crucial for avoiding persistent collisions in Multi-Armed Bandit (MAB) channel selection problems.
• Correlated Equilibrium: A more general concept where radios coordinate their actions based on a shared external signal, often achieving higher aggregate utility than a standard Nash Equilibrium. This models scenarios where a Spectrum Broker broadcasts a public coordination signal.