A Seminar report On

Gaming Theory in Routing Algorithms

Submitted By Devangini Patel

Guided By Priyanka Sharma

Concern Faculty Sapan Mankand

2008-09 Dept of Computer Engineering Institute of Technology Nirma University for Science and Technology

Certificate

Submitted towards the partial fulfillment of curriculum requirement of Semester VI of B. Tech. in Computer Engineering.

Guide

Priyanka Sharma

Acknowledgement

I am grateful to those who supported me during the seminar work which includes my guide, concerned faculty as well as all my batch mates. My guide always provided me with proper guidance of how to actually get about things and always co-operated with me during difficult times. But the response I received from the first presentation was the biggest factor which motivated me and made me think that I was on the correct path on understanding this topic and making people understand it as well as conveying its significance in the practical world. My concerned faculty also supported me and his remarks made me more confident about my work and it was the biggest factor involved in improving my performance. I am also thankful to my friends who let me know about my weak points during the presentation work so that I could I improve upon them later on.

Abstract

Gaming theory is such a branch of Applied Mathematics which is used in all spheres of life knowingly or unknowingly. The emergence of the internet has motivated the development of algorithms for finding equilibria in games, markets, computational auctions, peer-to-peer systems, and security and information markets. For many years, engineers have been trying to produce routing algorithms that are competitive enough to handle the maximum traffic load problems and situations like Congestion Reflection and Load Bargaining. The situation of routers in action is compared with Players Playing a Non-Cooperative game and the effects of Nashs Equilibrium on this situation are evaluated. The other heuristic methods used for analyzing and studying the routing algorithms are Artificial Intelligence, Inverse Game Theory better known as Algorithmic Mechanism Design and Economic Theory. The other application of gaming theory in the computer field include congestion control, dynamic topology formation in networks, efficient sharing of resources such as a CPU or Distributed Networks, overlay networks, wireless and sensor networks, security assessment and enhancement, fairness in forwarding and medium access, network pricing.

Index Chapter No.

1. 2. 2.1 2.2 2.2.1 2.2.2 3. 3.1 3.2 3.2.1 3.2.2 3.3 4. 4.1

Topic
Introduction Prerequisite Routing Routing Algortihms Link State Algorithms Distance Vector Alogrithms Gaming Theory

Page No.
7 8 8 8 9 10 13 13 14 15 15 15 18 18

Prescriptive or normative analysis Types of Games Extensive form Normal form Nashs equilibrium Gaming Theory and Routing Reason for introducing Gaming Theory in Routing

4.2 4.3

Visualizing routing as a game Parameters in routing with Game Theory

18 18

5. 6. 6.1 6.1.1 7. 8. 8.1 8.2 8.3 8.4 9. 10. 11. 12. 13.

Pareto Optimal Solution

19

Application of Gaming Theory in Routing 20 Elections Rules for the Election of Agents Co-operative load balancing Some simulation results Malicious mode of operation Occasional defects Occasional defects with more social nodes Practical situation Advantages Disadvantages Evolutionary Game Theory Conclusion References 22 22 25 27 27 28 29 30 31 32 33 34 35

1. Introduction
As defined by Bo Bernhardsson, Game Theory is the application of Multi Person Decision problems. Gaming theory is such a branch of Applied Mathematics which we use in our daily lives but we are unaware of the fact that we use gaming theory some way or the other. Gaming theory is used for analyzing situations and finding some optimal solution. The basic idea of gaming theory is to not think in terms of success of the main person but to think about all the players of the game and make them cooperate in such a way that both the players win in some sense. Gaming theory is also used to increase the probability of winning by suggesting certain moves or actions when we know or dont know the opponents moves (depending on the situation). Routing is a problem for all networks and it is the biggest factor affecting the price we pay for internet as the time we have to wait for the information to arrive at our PC. Gaming theory can be used to control congestion and to balance load, which are considered to be the well known problems of networking. These problems arise frequently in economics and also in engineering. The basic assumptions that underline the theory are that decisions makers are rational, and take into account their knowledge or expectations of other decision makers behavior. The questions studied in game theory often arise in applications in engineering, such as control theory, optimization, and information theory. Game theory gives tools to understand issues present when decision makers interact, such as the value of information and what is meant by rational and optimal strategies. One application in engineering is in design of control systems, where one agent, the controller tries to take into account the worst possible action of a hostile agent (disturbances). Other applications arise from sharing of common resources such as a CPU or Networks. There are connection to mathematics, optimization, mathematical statistics, information theory, control theory and economics. Over the times of science, game theory has found applications in various fields of a social human, but the realization of the application to computation has still been far from implementation primarily because of the fact that Game Theory in itself is a vast subject and it requires dedicated study and time for understanding, which programmers in computation either find too boring to study or as a waste of resources.

2. Prerequisite
2.1 Routing
Routing is the process of selecting paths in a network along which to send network traffic. Routing is performed for many kinds of networks, including the telephone network, electronic data networks (such as the Internet), and transportation networks. Routing used for transmission of logically addressed packets from their source toward their ultimate destination through intermediate nodes; typically hardware devices called routers, bridges, gateways, firewalls, or switches. General-purpose computers with multiple network cards can also forward packets and perform routing, though they are not specialized hardware and may suffer from limited performance. The routing process usually directs forwarding on the basis of routing tables which maintain a record of the routes to various network destinations. Thus, constructing routing tables, which are held in the routers' memory, is very important for efficient routing. Most routing algorithms use only one network path at a time, but multi path routing techniques enable the use of multiple alternative paths.

2.2 Routing Algorithms

Routers use routing algorithms to find the best route to a destination. BASED on how routers gather information about the structure of a network and their analysis of information to specify the best route, we have two major routing algorithms: global routing algorithms and decentralized routing algorithms. In decentralized routing algorithms, each router has information about the routers it is directly connected to -- it doesn't know about every router in the network. These algorithms are also known as DV (distance vector) algorithms. In global routing algorithms, every router has complete information about all other routers in the network and the traffic status

of the network. These algorithms are also known as LS (link state) algorithms.

2.2.1 Distance Vector Algorithms

This includes identifying the best route between two nodes of the network using Dijkstra shortest path algorithm. Here, a router, based on information that has been collected from other routers, builds a graph of the network. The Graph contains routers, links, and every link is labelled with a number called the weight or cost. This number is a function of delay time, average traffic, and sometimes simply the number of hops between nodes. If there are two links between a node and a destination, the router chooses the link with the lowest weight Here the distance between two routers can be found out as follows: 1. Identify the routers that are physically connected to them and get their IP addresses When a router starts working, it first sends a "HELLO" packet over network. Each router that receives this packet replies with a message that contains its IP address. 2. Measure the delay time (or any other important parameters of the network, such as average traffic) for neighbor routers In order to do that, routers send echo packets over the network. Every router that receives these packets replies with an echo reply packet. By dividing round trip time by 2, routers can count the delay time. (Round trip time is a measure of the current delay on a network, found by timing a packet bounced off some remote host.) Note that this time includes both transmission and processing times -- the time it takes the packets to reach the destination and the time it takes the receiver to process it and reply. 3. Broadcast its information over the network for other routers and receive the other routers' information In this step, all routers share their knowledge and broadcast their information to each other. In this way, every router can know the structure and status of the network.

10

Figure 1-graph of a network As shown in figure 1, there are five routers connected and a packet has to be transmitted from router A to router E. Now, using link state algorithm the path would come out to be Path A-B-D-E.

2.2.2 Link State Algorithm

Here, every router has a routing table that shows it the best route for any destination. It counts the weight of the links directly connected to it and saves the information to its table. In a specific period of time, it sends its table to its neighbouring routers (not to all routers) and receives the routing table of each of its neighbours. Based on the information in its neighbours' routing tables, it updates its own.

Figure 2-example of network As shown in figure 2, twelve routers have been connected. Suppose we have to send a packet from A to H then router A would contain the shortest routes for all the routers in its routing table and accordingly send the packet all on that path.

11

Destination A B C D E F G H I J K

Weight 8 20 28 20 17 30 18 12 10 0 6

Line A A I H I I H H I --K

L 15 K Table 1-A typical network graph and routing table for router J As the table 1, it shows, if router J wants to get packets to router D, it should send them to router H. When packets arrive at router H, it checks its own table and decides how to send the packets to D. In DV algorithms, each router has to follow these steps: 1. It counts the weight of the links directly connected to it and saves the information to its table. 2. In a specific period of time, it send its table to its neighbor routers (not to all routers) and receive the routing table of each of its neighbors. 3. Based on the information in its neighbors' routing tables, it updates its own. One of the most important problems with DV algorithms is called "count to infinity. "Imagine a network with a graph in figure 3. As you see in this graph, there is only one link between A and the other parts of the network.

Figure 3-example of link state

12

A A B C 0,1,B 2,B

B 1,A 0,1,C

C 2,B 2,C 0,-

D 3,C 3,D 1,C

D 3,B 2,C 1,D 0,Table 2-Network graph and routing tables Now imagine that the link between A and B is cut. At this time, B corrects its table. After a specific amount of time, routers exchange their tables, and so B receives C's routing table. Since C doesn't know what has happened to the link between A and B, it says that it has a link to A with the weight of 2 (1 for C to B, and 1 for B to A -- it doesn't know B has no link to A). B receives this table and thinks there is a separate link between C and A, so it corrects its table and changes infinity to 3 (1 for B to C, and 2 for C to A, as C said). Once again, routers exchange their tables. When C receives B's routing table, it sees that B has changed the weight of its link to A from 1 to 3, so C updates its table and changes the weight of the link to A to 4 (1 for C to B, and 3 for B to A, as B said). This process loops until all nodes find out that the weight of link to A is infinity. This situation is shown in the table below. In this way, experts say DV algorithms have a slow convergence rate. B Sum of weight to A after link cut Sum of weight to B after 1st updating Sum of weight to A after 2nd updating Sum of weight to A after 3rd updating Sum of weight to A after 4th updating Sum of weight to A after 5th updating Sum of weight to A after nth updating Table 3-"count to infinity" problem One way to solve this problem is for routers to send information only to the neighbors that are not exclusive links to the destination. For example, in this case, C shouldn't send any information to B about A, because B is the only way to A. ,A 3,C 3,C 5,C 5,C 7,C ... C 2,B 2,B 4,B 4,B 6,B 6,B ... D 3,C 3,C 3,C 5,C 5,C 7,C ...

13

3. Gaming Theory
Game theory is a branch of applied mathematics that is used in the social sciences (most notably economics), biology, engineering, political science, international relations, computer science, and philosophy. Game theory attempts to mathematically capture behavior in strategic situations, in which an individual's success in making choices depends on the choices of others. While initially developed to analyze competitions in which one individual does better at another's expense (zero sum games), it has been expanded to treat a wide class of interactions, which are classified according to several criteria. Today, "game theory is a sort of umbrella or 'unified field' theory for the rational side of social science, where 'social' is interpreted broadly, to include human as well as non-human players (computers, animals, plants)" (Aumann 1987). Traditional applications of game theory attempt to find equilibria in these games. In equilibrium, each player of the game has adopted a strategy that they are unlikely to change. Many equilibrium concepts have been developed (most famously the Nash equilibrium) in an attempt to capture this idea. These equilibrium concepts are motivated differently depending on the field of application, although they often overlap or coincide. This methodology is not without criticism, and debates continue over the appropriateness of particular equilibrium concepts, the appropriateness of equilibria altogether, and the usefulness of mathematical models more generally.

3.1 Prescriptive or normative analysis

Cooperate Defect Cooperate Defect -1, -1 0, -10 -10, 0 -5, -5

Figure 4-example of prisoners dilemma On the other hand, some scholars see game theory not as a predictive tool for the behavior of human beings, but as a suggestion for how people ought to behave. Since Nash equilibrium of a game constitutes one's best response to the actions of the other players,

14

playing a strategy that is part of Nash equilibrium seems appropriate. However, this use for game theory has also come under criticism. First, in some cases it is appropriate to play a non-equilibrium strategy if one expects others to play non-equilibrium strategies as well. For example, Guess 2/3 of the average. Second, the Prisoner's dilemma presents another potential counterexample. In the Prisoner's Dilemma, each player pursuing his own self-interest leads both players to be worse off than had they not pursued their own self-interests.

3.2 Types of Games

The games studied in game theory are well-defined mathematical objects. A game consists of a set of players, a set of moves (or strategies) available to those players, and a specification of payoffs for each combination of strategies. Most cooperative games are presented in the characteristic function form, while the extensive and the normal forms are used to define non cooperative games. Cooperative gaming is the type of gaming that makes people cooperate to win the game, they form allies and groups to play there games and eventually the grouping is the factor that makes there victory possible. This situation is directly derived from the real life examples. Like countries wage wars and form allies, similarly in the modeling strategy also players can wage wars and also form allies. Non Cooperative gaming is the counterpart of the cooperative gaming, in which players take each other player as there enemy and make moves so as to hamper the other players mobility and eventually win. This type of gaming is based on individualistic ideas, in which players optimize there own good put i.e. they always try to maximize there own profit.

15

3.2.1 Extensive form

Figure 5-An extensive form game The extensive form can be used to formalize games with some important order. Games here are often presented as trees as shown in figure 5. Here each vertex (or node) represents a point of choice for a player. The player is specified by a number listed by the vertex. The lines out of the vertex represent a possible action for that player. The payoffs are specified at the bottom of the tree. In the figure 5, there are two players. Player 1 moves first and chooses either F or U. Player 2 sees Player 1's move and then chooses A or R. Suppose that Player 1 chooses U and then Player 2 chooses A, then Player 1 gets 8 and Player 2 gets 2. The extensive form can also capture simultaneous-move games and games with imperfect information. To represent it, either a dotted line connects different vertices to represent them as being part of the same information set (i.e., the players do not know at which point they are), or a closed line is drawn around them.

3.2.2 Normal form

Player 2 chooses Left Player 1 chooses Up Player 1 chooses Down 4, 3

Player 2 chooses right 1, 1

0, 0

3, 4

Figure 6-Normal form or payoff matrix of a 2-player, 2-strategy game

16

The normal (or strategic form) game is usually represented by a matrix which shows the players, strategies, and payoffs (see the example to the right). More generally it can be represented by any function that associates a payoff for each player with every possible combination of actions. In figure 6, there are two players; one chooses the row and the other chooses the column. Each player has two strategies, which are specified by the number of rows and the number of columns. The payoffs are provided in the interior. The first number is the payoff received by the row player (Player 1); the second is the payoff for the column player (Player 2). Suppose that Player 1 plays Up and that Player 2 plays Left. Then Player 1 gets a payoff of 4, and Player 2 gets 3. When a game is presented in normal form, it is presumed that each player acts simultaneously or, at least, without knowing the actions of the other. If players have some information about the choices of other players, the game is usually presented in extensive form.

3.3 Nashs equilibrium

"When everyones playing their best move to everyone ELSES best move, no-ones going to move." The idea of this natural "sticking point" is that no single player can benefit from unilaterally changing his or her move -- a noncooperative best-response equilibrium. Competitive Markets come to rest at Nash equilibrium, and the special structure of competitive markets makes them efficient. But it is important to recognize that MOST Nash-Equilibria are NOT efficient. It's just the idea of getting the "whole pie" -- that if we're really using the whole pie, then no one can get any more unless someone else takes less. That's the economist's basic idea of allocative efficiency. The Nash Theorem maintains its focus on rivalries with mutual gain; a perceptual focus of Nash's mathematical vision found in the light of Leon Walrus' General Equilibrium Theory (published 1874) and John von Neumann's and Oskar Morgenstern's theory of games (1944), now simply called Game Theory. Nash later established his own idea of dominant strategy Equilibria through maximization solutions for zero-sum games. He did this with original mathematical

17

techniques to demonstrate the existence of methods for finding a measurable equilibrium in a general class of non-cooperative games. "The concept of a Nash equilibrium n-tuple is perhaps the most important idea in non cooperative game theory. Whether we are analyzing candidates' election strategies, the causes of war, agenda manipulation in legislatures, or the actions of interest groups, predictions about events reduce to a search for and description of Equilibria. Put simply, equilibrium strategies are the things that we predict about people." -P.Ordeshook. John Nash can be credited against astonishing odds with making a normative distinction between cooperative and noncooperative games, and for using mathematical models to support and exemplify his research. But his contribution to theory has become (neo-Nash) a web of explanation(s) and justification(s) far beyond the original conceptions of the author, and has also progressed steadily into the social world.

18

4. Gaming Theory and Routing

4.1 Reason for introducing Gaming Theory in Routing
Usually, what tends to happen is that routers go for selfish routing, in which users compete for resources and users decide path. The advantages include greater adaptability (in terms of change in network parameters) and it considers individual users profit. But this situation suffers from many problems which are: users must have sufficient knowledge, the performance of network is degraded (Price of Anarchy), traffic and congestion and lower bandwidth utility.

4.2 Visualizing routing as a game

Here, a router can be considered as a player and routing as a game. The task of each player is to choose the best path on which he can place his data to reach the destination. The available strategies are: co-operative or non-co-operative. Co-operative player can choose to cooperate with the other router and talk to it or form allies with it and share there information on choosing the most efficient path for channels effectively. Non co-operative can be enemies for fighting the best path to be used.

4.3 Parameters in routing with Game Theory

A three variable weight function can be used to derive the weight of any link for the gaming context. The three variables are as follows: 1. Reach: By reach we specify the closest reachability function in terms of Hop Count. This causes a few arguments but still the other two variables cover it up for the deficiency of this variable. 2. Activation: The probability that the sent data packet will reach its destination, for the amount of time the destination remains active and by what ratio does it accept or reject the data being sent from one end to another. 3. Connection: This variable specifies the bandwidth in the channel. This is a important variable as it covers the defects of the Reach variable, and causes the variable to FLOAT with optimization. So a function Weight (reach, activation, connection) is what defines the weight of the link in question.

19

5. Pareto Optimal Solution

Pareto Optimality is a factor that has become the de facto in differentiating real life problem from hypothetical ones. Consider the following example to better understand the concept.

Figure 7-example of rate allocation

The simple introductory sample shown in figure 7 portrays three sources and one destination. Each link has a Finite Capacity and the table on the right shows the different possible sending rates for the sources: All the rate allocation show in the table 7 above correspond to situations where given the maximum capacity of each link, no user can increase its emission rate without decreasing the rate of another user. Such a solution is called Pareto Optimum and is clearly desirable in computer networks. Another example might be a communication channel, which has a finite bandwidth of say 200 kbps. Here if one of the users has to increase its transfer rate in the case when the bandwidth is being fully utilized there can be no case when he can do so without decreasing the transfer rates of other users.

20

6. Application of Gaming Theory in Routing

There is a formation of agencies through elections and an agent is entirely uncommitted. Election is irrevocable for each specific playing of the game and election choices are expected to vary as the game is repeated. The election rule is that one agency elected during each stage of election. In a stage of elections, more than one of the players has voted to give his acceptance to another player to become his agent (and as if with "power of attorney") then a random process, or "chance move", determines only one of those votes as effective, and one of the players has elected one of the other players to be agent for him. It takes n-1 steps of effective election to achieve the final election of an agency for the realization of "Pareto Efficiency", it is needful that the grand coalition level is attained. Here it is considered as a combination of both co-operative and non co-operative games. Agencies is considered as the means for the reduction of the potentially quite general concept of a coalition to a concept well suited for studies based on analysis of the players' motivations from the noncooperative viewpoint of separate and independent utility measures also that there was a connection with the studies on the topic of "the evolution of cooperation" (Axelrod et al). Instead of there being unlimited means by which coalitions might actually be formed or dissolved, and reformed, we have an election procedure through which any player may elect to accept any other player as his agent. In the context of studying a repeated game, this election process is such that the agent is entirely uncommitted and his election is irrevocable for each specific playing of the game and the election choices are expected to vary as the game is repeated. A set of rules can be devised so that there are election stages in each of which all of the players remaining independent (not represented by another player as agent) have, each of them, the option of electing another player as an accepted agent. It is natural for these rules to require convergence so that no more than (n-1) stages of election will ever be needed for a game of n players. Election rules need to do something to resolve the impasse of a situation where A votes to accept B as his agent but B simultaneously also votes similarly for A.

21

It is desirable to allow elections to be repeated when an election had failed to result in the election of any agency power, which suggests that election rules which had the effect that at most one agency could be elected at any stage of the election process would be most convenient. Concerning the general concept of transforming a cooperative game into a form where all cooperation must be realized by means of the election of agencies, it can be remarked that this is analogous to thinking of committees as being such that all effective actions of a committee must take the form of an action by the "chairperson" of the committee. If one begins with a quite general CF game and then if one introduces a game formally requiring that all coalition benefits must be attained through the means of the action of agents who have the authority to represent all the members of the coalition then the "agencies game" resulting from this still has the same validly derivable characteristic function as the original game. In essence the coalitions simply have the same potentialities as before, for these potentialities to be exploited, the members of a coalition would need to conspire on a practical procedure for electing agents successively and finally taking the effective action of the coalition as an action of the agent finally elected to represent all of the members of that coalition. The Currently Applied Election Rules for Agencies was initially used just for the purpose of simplifying the format of model games and simplifying the calculations needed to find the corresponding equilibria. But in the process it was found that a simplifying format of election rules seems to be adequate and convenient. If in a stage of elections more than one of the players has voted to give his acceptance to another player to become his agent (and as if with "power of attorney") then a random process, or "chance move", determines only one of those votes as effective, and one of the players has elected one of the other players to be agent for him. The repeated election opportunity was to be given only with probability of (1-e4) (one minus epsilonsub4). Good results can be obtained mathematically by considering the limiting results as e4 (epsilonsub4) tended to the limit of zero. In general, if there are n players of a game, with this procedure, it would take n-1 steps of effective election to achieve the final election of an agency for the "grand coalition" (which is the coalition including all of the players of the game). And in general, for the realization of "Pareto Efficiency", it is needful that the grand coalition level is attained.

22

6.1 Elections
"Reduction of Coalitions to Agencies, a Scheme for the Analysis Of Cooperative Games"

6.1.1 Rules for the Election of Agents

The following rules are to be implemented alone or as a combination of few. These rules had been devised for practical purposes. At the beginning of the election process, when none of the players have yet become agents, each player can vote to accept a specific other player as his agent, or he can decline to make any such offer and then he thus effectively votes for himself. This election rule was used for some early calculations of model game strategies and was designed to avoid a possible ambiguity by disqualifying both votes if player A voted for player B and player B simultaneously voted for player A. With just two players that rule of disqualification simply led to a repeat election where the voting might go differently and A or B could become elected as the sole agent and as representing both of them. And the analogous more general possible election outcome, with n players, would be that a "cycle" of players is elected as approved for agency functions. That is, for example, if A votes to approve of B and if B, C, and D vote similarly to approve of (respectively) C, D, and A, then we have that the "cycle" (A,B,C,D) has been elected. We can resolve ambiguity here appropriately by awarding the agency power randomly with the probability of 1/4 for each of the four players in the "cycle" and with the elected agent representing the total ensemble of four. It is natural to have it specified as a rule that election cannot be declined. This is natural because by being elected as an agent a player acquires simply enhanced strategic powers; he has all of his original strategic options of the normal form presentation of the game plus also all of those of any of the other players for whom he has become agent.

23

"Chain election rule" accelerates the election process as if by anticipating the results of repeated voting. If A votes for D and simultaneously D votes for F while F votes for himself then we declare that F is elected to be agent for both D and A, with of course also the representation of himself too. This is the rule of "chain sequence election". It is in the nature of agency that if D had become the agent of A and then after that D gave agency power to F then F would become the agent for both A and D. So this rule accelerates the coalescence process. So a "solo" is elected if one player votes for himself as his accepted agent, that is, he elects himself if he declines to vote for any other player as acceptable to be agent for him. On the other hand, a "cycle" is elected if there is a cyclical sequence of n voting players each of which has voted for the next member of the sequence. And a "solo" can also be viewed as a 1-cycle, thus a cycle with only one member. Then the "chain sequence election" rule may result in the election of either a "solo" or a "cycle" to exercise agency for all of the members of a chain of electors. Such a chain is a linearly ordered sequence where each member has voted for the next except for the last member who is either a "solo" or a member of a "cycle". Then the cycle including the last member of the linear chain is elected to have agency for all of the members of that chain. And it may of course be that that "cycle" is just a 1-cycle or "solo" which is the not unlikely case when the last member of the "chain" has simply voted for him. And then the cycle elected to exercise agency would become the random choice of one of its members. Thus, if there were required another stage of elections, then that randomly chosen member of the elected cycle would be the only player surviving, of the chain and the cycle, to continue with the process of coalescence through the election of agencies. There is an interesting alternative approach that seems very natural in the case of the election of a "cycle". The alternative is to introduce a sort of "robotic attorney" who takes over the function of agency when a cycle (of more than one member) has been elected. This could seem to be very complicating since it does amount to introducing a new player into the game. But it isn't complicating in a practical sense that is for the calculation of solutions with game models, because the strategic options of the newly introduced "robotic" or "virtual" players replace options that would otherwise be those of original players.

24

It is also simplifying that the robotic player can be assigned a very simple payoff function so that his payoff is simply measured by the probability of his being employed. Thus he is like a corporation lawyer working on a basis of lawyer's hours for payment. No matter what the gains of the actual original players are the payoff of this sort of robot agent is to be calculated as simply the probability that he has a role (as an agent). Thus most simply, if the game were one of 2 players, the robotic player would be introduced ONLY when the cycle (1,2) of the two players had been elected. So the probability of the employment of the robotic attorney would be that of the event that each of the original players votes to accept the other as agent. So the robotic player is simply motivated, in Model calculations, to maximize the product of the two acceptance probabilities. And that leads the robot, effectively, to cultivate a reputation of reasonably favoring the interests of both of the original players. This alternative approach to cycles functions similarly to the other rule, that of the random choice of a member of the cycle to become the agent, but when the Pareto boundary of the game is convex it tends to be naturally more efficient in the Pareto efficiency sense. The result of this election process is a group of clusters of routers as shown in figure 8. A Cluster is a set of routers with n routers in it, where each router selects its cluster for computation on its own on a fixed set of election rules.

Figure 8-clusters formed as a result of elections

25

7. Co-operative load balancing

There is another concept called cooperative load balancing which can be derived from the fundamentals of Pro-Cooperative Gaming. In the figure 7, the rate allocation was shown in the table corresponding to situations where given the maximum capacity of each link, no user can increase its emission rate without decreasing the rate of another user. In general there exists infinite Pareto optimum solutions for a given problem. Therefore, additional constraints must be introduced. By inspection of the table of figure 7, it can be seen that the rate allocation corresponding to line 2 and 4 yield a total link utilization which is maximum. However, these allocations prevent S3 from accessing the network. Clearly, the first line if the table gives the most attractive solution because it gives a fair share of the network resources to each user. Fairness between users is very important and has to be considered when selecting Pareto-optimum solution. In practice, the problem of rate allocation is worsened by the fact that the different sending rate are not distributed by the network but are instead selected by each user who may react to others decisions. According to the pro cooperative game theory there seems to be the possibility that theory for cooperative games may work better for games which tend to strongly reward cooperating of the players that for those which could be regarded as tending more towards favoring non-cooperative behavior of the players. A game of the sort that most simply favors cooperation is a game which approximates to a maximally simple game of bargaining. If agreement of all players in a three-person game is necessary for them to receive the main payoff total and coalitions of merely two of the players could get only comparatively quite small amounts then the game intrinsically favors the cooperation needed for the realization of the benefits of the "grand coalition. Cooperation is not always intrinsically favored, in nature, or in human affairs. Sporting events would become absurd if Sumo wrestlers were to spend ALL of their time in polite ceremonies and respectful bows. And Nature allows the evolution of parasitism and predation as well as the evolution of symbiotic relationships.

26

These relations do definitely derive that we can make routers cooperate and do load balancing by rewarding there agent by taking his load on there shoulders. One good example of this in nature would be the relationship between pea and rhizobia. The rhizobia live on the roots of pea plants and provide them with nitrogen and the pea plants in return provide them with food.

Nitrogen Pea Food Figure 9-relationship between pea and rhizobia Rhizobia

27

8. Some simulation results

8.1 Malicious mode of operation

Figure 10-Graph of average payoff versus du fraction in malicious mode. In malicious mode, the routers will be tempted to become noncooperative. Here, du represents the defected nodes (non-cooperative nodes) and snr represents the social nodes. Here, there are 1024 nodes and the average nodes along any routing path are 5.5. It can be seen from figure 9 that the payoff of defected nodes is almost zero whereas the payoff of social nodes is more when they are in majority. It can also be concluded that the payoff of social node is negative when more nodes are non-cooperative.

28

8.2 Occasional defects

Figure 11-graph of number of players versus total payoff when 10% of the total nodes are cooperating Here, if the opposite router is innocent then the router should cooperate with probability p or defect with probability p where p is usually taken as 0.2. In this case occasional defects are detected. The random situation supports less number of players because it can only support some maximum routers which do not cooperate whereas compared to this situation the social situation supports more number of maximum players because in this case at least some player are cooperating due to which some non-cooperative players can be introduced to get the same payoff. In the random situation, the payoff is almost constant and it does not increase much with the increase with the number of players. But in the social mode, the payoff increases as the number of players increases because the number of cooperating routers also increases.

29

8.3 Occasional defects with more social nodes

Figure 12- Figure 11-graph of number of players versus total payoff when 20% of the total nodes are cooperating As compared to the previous example, the payoff would definitely increase because the number of cooperating routers has increased for the same number of total routers.

30

8.4 Practical situation

Figure 12-graph of number of players versus total payoff for a practical situation

In a practical situation a router may not cooperate at any random time (router may go down). So in such cases, for some time the number of players which would get a good payoff decreases because many routers which route through these routers depend upon its state. But in some time, it would go up and the situation would return to the social norm as shown in figure 12.

31

9. Advantages

This form of routing requires no or little up gradation and is most efficient. This is the best way to start standardization process. It is easy to calculate the efficiency because some part of the routers memory has to be reserved to store information about the elections. It is easy to simulate the results by softwares like Mat lab, Mathematica, OMNET++, C++, CNet, and OMNEST. It also has feature like load balancing which is one of the major problems of networking and most important of all it satisfies the Pareto Optimal solution. Gaming theory can be seen as a way of providing fairness in any routing algorithm and thus can be used as a way of improving the overall efficiency of the routing algorithm.

32

10. Disadvantages

The implementation of gaming theory can not yield as good results as in selfish routing for an individual player in situations as Internet. It so adaptive in terms of topology changes because when the topology changes and the routing has to be done as per the topology then the game has to be played again. Here, we cannot give priorities properly because there would be certain routers which are more powerful (have more buffer space, memory space) and in such cases it would be desirable for these routers to be the agents. Selfish routing better than routing with game theory in situations like INTERNET Selfish routing is no worst than the optimal routing of twice as much traffic for any cost function, network topology and traffic matrix.

33

11. Evolutionary Game Theory

An area of more recent research relating to game theory is the work of the theoretical biologists who have studied the evolution of cooperation among species. The specific mathematical game that they have most frequently studied is the so-called "Prisoner's Dilemma" game. But they have studied it, effectively, as a cooperative game by considering it in the context of repeated play, the game being replayed indefinitely many times. Under the conditions of repetition of the game it becomes non-cooperatively rational for a player to arrive at an effectively cooperative accommodation with the other player. These studies have also involved modeling the players as relatively simple robots employing a limited variety of relatively simple strategies. These researches of the theoretical biologists seem to point to a way that game theory and game theorists can follow, a way that could have the value of avoiding arbitrary mathematical concepts or arbitrary rules for debating concerning imputations and coalitions.

34

12. Conclusion

The current industrial standard have enough space to accommodate and update their existing systems to the theory that we propose. Zero up gradation is required to implement this strategy. It has been observed by certain professionals that tere are improvements to the order of 64.2% in fully loaded cognito networks. An industrial study has shown the actualization costs of the plan and also propose a range of low to high end solutions which can be enforced on the older and existing routing infrastructure with minimal or no upgrades. With the advent of mobility in everyday networking, people are looking at standards in the form of data packets that need to be transferred. Everywhere throughout the networking scenario people need to take care of the future needs that will very soon arise in the form of data transparencies in the data packets standards in the industry. To accommodate and standardize this data people need to standardize strategies in the form of data transformation and game theory has always proven to be the most efficient way of data transfer. This is definitely not the only way around the current problems, but looking at the up gradation costs required this can be definitely called the most efficient way to start the standardization process.

35

13. References
1. Wikipedia.com 2. HowStuffWorks.com 3. A Path Breaking theory in Networking By Gunish Rai Chawla and Deveena Verma 4. Applications of Game Theory to Network Routing by A. Orda, R. Rom and N. Shimkin 5. Gaming Theory and Networking by Daniel Figueiredo 6. Gaming theory for networks by Christina 7. Selfish users in ALOHA by Dandan