Browsing by Subject "Game Theory"
Now showing 1 - 5 of 5
Results Per Page
Sort Options
Item A Game Theory Based Model of Human Driving with Application to Autonomous and Mixed Driving(2014-08-04) Yoo, Je HongIn this work, I consider the development of a driver model to better understand human drivers? various behaviors in the upcoming mixed situation of human drivers and autonomous vehicles. For this, my current effort focuses on modeling the driver?s decisions and corresponding driving behaviors. First, I study an individual driver?s reasoning process through game theoretic investigation. The driver decision model is modeled as the Stackelberg game, which is based on the backward information propagation. In the driver decision model, I focus on the drivers? insensible desires and corresponding unwanted traffic situations. With the comparison of the model and the field data, it is shown that the model reproduces the relationship between the driver?s inattentiveness and collisions in the real world. Next, the driving behavior control is presented. I propose a human-like predictive perception model of potential collision with an adjacent vehicle. The model is based on hybrid systematic approach. In turn, with the predictive perceptions, a driving safety controller is designed based on model predictive control. The model shows adequate predictive responses against the other vehicles with respect to the driver?s rationality. In sum, I present a driver model that corresponds to and predicts traffic situations according to a human driver?s irrationality factor. This model shows a meaningful similarity to the real-world crashes and predictive behaviors according to the driver?s irrationality.Item A Packet Scheduling Mechanism for Wireless Peer-to-Peer Content Distribution(2013-11-26) Liu, YaoThis thesis studies the problem of content distribution in wireless peer-to-peer networks with selfish nodes. In this problem a group of wireless nodes need to exchange a set of files over a lossless broadcast channel. Each node aims to maximize its own download rate and minimize its upload rate. We propose a distributed protocol that provides incentives for selfish nodes to participate in the content exchange. Our protocol does not require any exchange of money and reputation and hence can be easily implemented without additional infrastructure. Then, we will analyze the performance of our protocol by focusing on the import\-ant case in which the system contains two files that need to be distributed. We derive a closed-form expression of Nash Equilibrium and characterize the corresponding system performance in discrete time. Furthermore, we propose a distributed mechanism where the strategy of each node is only based on the observed history of the system and not on the private information of other nodes. We also study the performance characteristics of the systems that employ network coding to facilitate data exchange. We show that, due to the free rider problem network coding does not necessary improve the performance of the system and, in some cases, may lead to worse system performance. We propose a novel approach to this problem based on random coding. The performance of the network coding algorithms is validated by performing extensive simulation study.Item Inductive Causation on Strategic Behavior: The Case of Retailer and Manufacturer Pricing(2011-02-22) Fraire Dominguez, FranciscoModels of strategic behavior are usually too complex to conduct large scale analyses, and frequently rely on accurate descriptions of the strategic environment, or unrealistic assumptions which render empirical studies very sensitive to misspecification. This dissertation relates game-theoretic frameworks to models of causality inference and thus provides a reliable method to identify price leadership. Therefore, causal models can be used to study large sets of data without imposing strategic behavior a priori. A case study is provided by analyzing the supply chain relationship among Dominick's Finer Foods and its suppliers. Although our data required aggregation, this empirical analysis successfully determined causal patterns for 60 percent of our sample. Of these price leaderships, 70 percent elicit Manufacturer Stackelberg relationships which tend to be associated with manufacturers that hold big market shares, 25 percent elicit Retailer Stackelbergs which seem to be associated with the biggest retailer margin profits, and only 5 percent elicit a monopolistic retailer with vertical coordination. These results agree with observations made by other authors and the market structure of the 1990's. Moreover, the strategic relationship among the suppliers is also studied. Interestingly, the dominant firms tend to isolate themselves from the price leadership, whereas the second largest firms seem to become price leaders. Our studies agree with the market literature as well. In particular, we find price leadership in a firm which was identified as a low cost leader. Finally, we discovered that the private label does not lead any firm's price unless this firm is the provider of a generic brand.Item Noncooperative Games for Autonomous Consumer Load Balancing Over Smart Grid(2011-10-21) Agarwal, TarunTraditionally, most consumers of electricity pay for their consumption according to a fixed-rate. The few existing implementations of real time pricing have been restricted to large industrial consumers, where the benefits could justify the high implementation cost. With the advancement of Smart Grid technologies, large scale implementation of variable-rate metering will be more practical. Consumers will be able to control their electricity consumption in an automated fashion, where one possible scheme is to have each individual maximize their own utility as a noncooperative game. In this thesis, noncooperative games are formulated among the consumers of Smart Grid with two real-time pricing schemes, where the Nash equilibrium operation points are investigated for their uniqueness and load balancing properties. The first pricing scheme charges a price according to the average cost of electricity borne by the retailer and the second charges according to a time-variant increasing-block price. The zero revenue model and the constant revenue rate model, are the two revenue models being considered. The relationship between these games and certain congestion games, known as atomic flow games from the computer networking community, is demonstrated. It is shown that the proposed noncooperative game formulation falls under the class of atomic splittable flow games. It is shown that the Nash equilibrium exists for four different cases, with different pricing schemes and revenue models, and is shown to be unique for three of the cases, under certain conditions. It is shown that both pricing schemes lead to similar electricity loading patterns when consumers are interested only in the minimization of electricity costs. Finally, the conditions under which the increasing-block pricing scheme is preferred over the average cost based pricing scheme are discussed.Item Reverse Auction Bidding Further Elements to the Game Theory(2014-06-03) Li, JiaxingReverse Auction Bidding systems are increasingly used by some large corporations for the supply of buildings, an example is the major firm Target. The belief is that the Reverse Auction Bidding system improves the efficiency of the bidding system and leads to cost savings during the construction process. Neither statement has been shown to be correct at this time. A game theory was developed for the Reverse Auction Bidding system; this theory postulated that two sub-games exist within the overall Reverse Auction Bidding game. The first sub-game is between the purchaser and the set of bidders. The purchaser is presented with a group of lowest prices that under the rules of the game must be accepted. This group of prices has been shown to have a non-normal distribution in prior research at TAMU. If economic efficiency was to be maintained by the bidding system, one would expect a normal distribution with a tight range on the standard deviation, which does not occur. The second sub-game is between the bidders, who make use of the non-normal aspects of price group to maximize individual returns. All things being equal and given the intent of the game, the purchaser would expect the bidders return to be normally distributed with a small standard deviation representing a tight control on price, which has never been observed in game play. Three types of bidders have been postulated for the set, the first is an economically efficient bidder, an economically inefficient bidder, and a middle of the road bidder. This study aims to compare statistically the difference between economically efficient bidders, Type ? bidder, and economically inefficient bidders, Type ? bidder, in terms of the statistical properties of the return data. The central hypothesis is that a statistically evident bias exists between the average return generated by the Type ? bidder and the Type ? bidder. The addition of the two distributions along with the average return generated by a Type ? bidder results in the observed distribution for the group, L. The secondary hypothesis is that Type ? bidders minimize the price reduction for each bid. The first hypothesis is true, the Type ? bidder earn on average twice the returns of the Type ? bidder. The second hypothesis is not true, the Type ? bidder as a set do not attempt to minimize the bid differentials. Further research is suggested on the statistical properties of the bid differentials as more games are played at TAMU.