An LP Approach for Solving Two-Player Zero-Sum Repeated Bayesian Games

Abstract

This paper studies two-player zero-sum repeated Bayesian games in which every player has a private type that is unknown to the other player, and the initial probability of the type of every player is publicly known. The types of players are independently chosen according to the initial probabilities, and are kept the same all through the game. At every stage, players simultaneously choose actions, and announce their actions publicly. For finite horizon cases, an explicit linear program is provided to compute players' security strategies. Moreover, based on the existing results in [1], this paper shows that a player's sufficient statistics, which is independent of the strategy of the other player, consists of the belief over the player's own type, the regret with respect to the other player's type, and the stage. Explicit linear programs are provided to compute the initial regrets, and the security strategies that only depends on the sufficient statistics. For discounted cases, following the same idea in the finite horizon, this paper shows that a player's sufficient statistics consists of the belief of the player's own type and the anti-discounted regret with respect to the other player's type. Besides, an approximated security strategy depending on the sufficient statistics is provided, and an explicit linear program to compute the approximated security strategy is given. This paper also obtains a bound on the performance difference between the approximated security strategy and the security strategy.

Defense against unknown broadband jammer for time‐critical operation in smart grid

“…As a result, the complexity of this solution approach will increase as the number of stages increases. Therefore, motivated by Li et al [16], in the following proposition we will modify , such that player I needs to solve a single LP to find its NE strategy for T ‐stage repeated the game. Proposition 2 The LP to be solved by Player I at the start of T‐stage game is given by such that …”

“…Further, the end‐to‐end delay experienced by the transmitted messages depends on the number of times jammer successfully jams the transmission. In this regard, repeated non‐cooperative games are used to model various competitive scenarios in communication networks, where the nodes interact repeatedly over time [15–18]. A two‐player general sum repeated game, formulated in [15], illustrates how a smart jammer that acts either as a cheater to obtain more network resource or a saboteur to cause damage, can impact the LTE network.…”

“…A two‐player general sum repeated game, formulated in [15], illustrates how a smart jammer that acts either as a cheater to obtain more network resource or a saboteur to cause damage, can impact the LTE network. In [16], a repeated game is used to analyse underwater jamming. In [17, 18], the repeated interaction between buyers and sellers is modelled as a repeated auction.…”

A Reinforcement-learning-based Bidding Strategy for Power Suppliers with Limited Information

“…The contributions of this paper are outlined below: 1) This paper defines the bidding procedure of power suppliers with thermal power units as a repeated game [18] rather than the widely used Markov games, avoiding the stringent requirements on state transition.…”

Worst-Case Analysis for a Leader-follower Partially Observable Stochastic Game

“…Repeated Stackelberg games, repeated Bayesian games, and multi-period stochastic games were developed in order to capture the dynamic interaction between a leader and its adversary. Li et al (2018) analyzed and computed the agents' security strategies for a two-player zero-sum repeated Bayesian game. Examples of how to employ repeated Stackelberg games to model the interaction between a defender (the leader) and its adversary (the follower) can be found in the context of wildlife security (Yang et al 2014) and fisheries security (Haskell et al 2014).…”